전체 글

21. 서울시 범죄율 분석 || MinMaxscalimg 2021.11.24
20. 서울시 인구분석 || 다중회귀 2021.11.23
19. 세계음주데이터2 2021.11.23
제2회 성남여성 창업아이디어 경진대회 수상 2021.11.17
39일차 - 논문 수업 (GAN) 2021.11.16
38일차 - Style Transfer 2021.11.16
37일차 - 논문 수업 (Segmentation) 2021.11.16
36일차 - 논문 수업 (Segmentation) 2021.11.16
18. 세계음주 데이터 분석 2021.11.03
17. 서울 기온 분석 2021.11.02

21. 서울시 범죄율 분석 || MinMaxscalimg

2021. 11. 24. 13:24

728x90

# 구별 범죄율 분석 import numpy as np import pandas as pd crime_police = pd.read_csv('02. crime_in_Seoul.csv', thousands = ',', encoding = 'euc-kr') crime_police.head() 관서명 살인 발생 살인 검거 강도 발생 강도 검거 강간 발생 강간 검거 절도 발생 절도 검거 폭력 발생 폭력 검거 0 중부서 2 2 3 2 105 65 1395 477 1355 1170 1 종로서 3 3 6 5 115 98 1070 413 1278 1070 2 남대문서 1 0 6 4 65 46 1153 382 869 794 3 서대문서 2 2 5 4 154 124 1812 738 2056 1711 4 혜화서 3 2 5 4 96 63 1114 424 1015 861

crime_police.info() <class 'pandas.core.frame.DataFrame'> RangeIndex: 31 entries, 0 to 30 Data columns (total 11 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 관서명 31 non-null object 1 살인 발생 31 non-null int64 2 살인 검거 31 non-null int64 3 강도 발생 31 non-null int64 4 강도 검거 31 non-null int64 5 강간 발생 31 non-null int64 6 강간 검거 31 non-null int64 7 절도 발생 31 non-null int64 8 절도 검거 31 non-null int64 9 폭력 발생 31 non-null int64 10 폭력 검거 31 non-null int64 dtypes: int64(10), object(1) memory usage: 2.8+ KB

police_state = pd.read_csv('경찰청_경찰관서 위치, 주소_20200409.csv', thousands = ',', encoding = 'euc-kr') police_state.head() 청 서 지구대파출소 X좌표 Y좌표 주소 0 강원청 강릉경찰서 강동파출소 128.978300 37.727760 강원도 강릉시 강동면 안인리 764-1 1 강원청 강릉경찰서 강릉경찰서 128.906763 37.768700 강릉시 포남동 1113 2 강원청 강릉경찰서 남부지구대 128.897125 37.748968 강릉시 노암동 715-16 3 강원청 강릉경찰서 동부지구대 128.926315 37.774032 강릉시 송정동 740-3 4 강원청 강릉경찰서 북부지구대 128.875237 37.835265 강릉시 주문진읍 주문리 312-7

police_state[police_state['지구대파출소'] == '서울중부경찰서'] 청 서 지구대파출소 X좌표 Y좌표 주소 1464 서울청 서울중부경찰서 서울중부경찰서 126.989614 37.563514 중구 저동2가 62-1

police_seoul = police_state[police_state['청'] == '서울청'] police_seoul.head() 청 서 지구대파출소 X좌표 Y좌표 주소 1207 서울청 서울강남경찰서 논현1파출소 127.029316 37.513791 강남구 논현동 58-13 1208 서울청 서울강남경찰서 논현2파출소 127.033875 37.515220 강남구 논현동 89-13 1209 서울청 서울강남경찰서 삼성1파출소 127.060440 37.514852 강남구 삼성동 107-3 1210 서울청 서울강남경찰서 삼성2파출소 127.048063 37.511289 강남구 삼성동 114-6 1211 서울청 서울강남경찰서 서울강남경찰서 127.067177 37.509036 강남구 대치동 998

# 경찰서의 이름을 서울xxx경찰서 형태 # 중부서 > 중부경찰서 police_name = [] for name in crime_police['관서명'] : police_name.append('서울' + str(name[:-1]) + '경찰서') print(police_name) ['서울중부경찰서', '서울종로경찰서', '서울남대문경찰서', '서울서대문경찰서', '서울혜화경찰서', '서울용산경찰서', '서울성북경찰서', '서울동대문경찰서', '서울마포경찰서', '서울영등포경찰서', '서울성동경찰서', '서울동작경찰서', '서울광진경찰서', '서울서부경찰서', '서울강북경찰서', '서울금천경찰서', '서울중랑경찰서', '서울강남경찰서', '서울관악경찰서', '서울강서경찰서', '서울강동경찰서', '서울종암경찰서', '서울구로경찰서', '서울서초경찰서', '서울양천경찰서', '서울송파경찰서', '서울노원경찰서', '서울방배경찰서', '서울은평경찰서', '서울도봉경찰서', '서울수서경찰서']

# 경찰서의 구이름을 리스트로 출력 police_seoul police_address = [] for name in police_name : select_police = police_seoul[ police_seoul['지구대파출소'] == name ] # print(select_police['주소']) police_address.append(select_police.loc[:,'주소']) police_address [1464 중구 저동2가 62-1 Name: 주소, dtype: object, 1439 종로구 경운동 90-18 Name: 주소, dtype: object, 1286 중구 남대문로5가 561 Name: 주소, dtype: object, 1345 서대문구 미근동 165 Name: 주소, dtype: object, 1475 종로구 인의동 48-57 Name: 주소, dtype: object, 1421 용산구 원효로1가 12-12 Name: 주소, dtype: object, 1377 성북구 삼선동5가 301 Name: 주소, dtype: object, 1310 동대문구 청량1리동 229 Name: 주소, dtype: object, 1333 마포구 아현동 618-1 Name: 주소, dtype: object, 1413 영등포구 당산동3가 2-11 Name: 주소, dtype: object, 1366 성동구 행당동 192-8 Name: 주소, dtype: object, 1326 동작구 노량진1동 72 Name: 주소, dtype: object, 1259 광진구 구의동 254-32 Name: 주소, dtype: object, 1353 은평구 녹번동 177-15 Name: 주소, dtype: object, 1229 강북구 번1동 415-15 Name: 주소, dtype: object, 1282 금천구 신림8동 544 Name: 주소, dtype: object, 1460 중랑구 묵2동 249-2 Name: 주소, dtype: object, 1211 강남구 대치동 998 Name: 주소, dtype: object, 1253 관악구 봉천4동 산177-3 Name: 주소, dtype: object, 1242 강서구 화곡6동 980-27 Name: 주소, dtype: object, 1222 강동구 성내1동 540-1 Name: 주소, dtype: object, 1449 성북구 종암1동 3-1260 Name: 주소, dtype: object, 1273 구로구 구로2동 436 Name: 주소, dtype: object, 1359 서초구 서초3동 1726 Name: 주소, dtype: object, 1402 양천구 신정6동 321 Name: 주소, dtype: object, 1388 송파구 가락본동 9 Name: 주소, dtype: object, 1297 노원구 하계동 250 Name: 주소, dtype: object, 1341 서초구 방배동 455-10 Name: 주소, dtype: object, 1431 은평구 불광2동 산24 Name: 주소, dtype: object, 1303 도봉구 창4동 17 Name: 주소, dtype: object, 1397 강남구 개포동 14 Name: 주소, dtype: object]

gu_name = [] for name in police_address : tmp = name.str.split() tmp = tmp.tolist() tmp_gu = tmp[0] gu_name.append(tmp_gu[0]) gu_name ['중구', '종로구', '중구', '서대문구', '종로구', '용산구', '성북구', '동대문구', '마포구', '영등포구', '성동구', '동작구', '광진구', '은평구', '강북구', '금천구', '중랑구', '강남구', '관악구', '강서구', '강동구', '성북구', '구로구', '서초구', '양천구', '송파구', '노원구', '서초구', '은평구', '도봉구', '강남구']

crime_police['구별'] = gu_name crime_police.head() 관서명 살인 발생 살인 검거 강도 발생 강도 검거 강간 발생 강간 검거 절도 발생 절도 검거 폭력 발생 폭력 검거 구별 0 중부서 2 2 3 2 105 65 1395 477 1355 1170 중구 1 종로서 3 3 6 5 115 98 1070 413 1278 1070 종로구 2 남대문서 1 0 6 4 65 46 1153 382 869 794 중구 3 서대문서 2 2 5 4 154 124 1812 738 2056 1711 서대문구 4 혜화서 3 2 5 4 96 63 1114 424 1015 861 종로구

# 구별 범죄건수들의 합계 구하기 crime_police.groupby('구별').sum() crime_sum = pd.pivot_table(crime_police, index = '구별', aggfunc=np.sum) crime_sum 강간 검거 강간 발생 강도 검거 강도 발생 살인 검거 살인 발생 절도 검거 절도 발생 폭력 검거 폭력 발생 구별 강남구 349 449 18 21 10 13 1650 3850 3705 4284 강동구 123 156 8 6 3 4 789 2366 2248 2712 강북구 126 153 13 14 8 7 618 1434 2348 2649 강서구 191 262 13 13 8 7 1260 2096 2718 3207 관악구 221 320 14 12 8 9 827 2706 2642 3298 광진구 220 240 26 14 4 4 1277 3026 2180 2625 구로구 164 281 11 15 6 8 889 2335 2432 3007 금천구 122 151 6 6 4 3 888 1567 1776 2054 노원구 121 197 7 7 10 10 801 2193 2329 2723 도봉구 106 102 10 9 3 3 478 1063 1303 1487 동대문구 146 173 13 13 5 5 814 1981 2227 2548 동작구 139 285 5 9 5 5 661 1865 1587 1910 마포구 247 294 10 14 8 8 813 2555 2519 2983 서대문구 124 154 4 5 2 2 738 1812 1711 2056 서초구 249 393 6 9 6 8 1091 2635 2098 2399 성동구 119 126 8 9 4 4 597 1607 1395 1612 성북구 124 150 4 5 5 5 741 1785 1855 2209 송파구 178 220 10 13 10 11 1129 3239 2786 3295 양천구 105 120 3 6 5 3 672 1890 2030 2509 영등포구 183 295 20 22 12 14 978 2964 2961 3572 용산구 173 194 14 14 5 5 587 1557 1704 2050 은평구 141 166 6 9 3 3 711 1914 2306 2653 종로구 161 211 9 11 5 6 837 2184 1931 2293 중구 111 170 6 9 2 3 859 2548 1964 2224 중랑구 148 187 9 11 12 13 829 2135 2407 2847

crime_sum['강간검거율'] = crime_sum['강간 검거']/crime_sum['강간 발생']*100 crime_sum['강도검거율'] = crime_sum['강도 검거']/crime_sum['강도 발생']*100 crime_sum['살인검거율'] = crime_sum['살인 검거']/crime_sum['살인 발생']*100 crime_sum['절도검거율'] = crime_sum['절도 검거']/crime_sum['절도 발생']*100 crime_sum['폭력검거율'] = crime_sum['폭력 검거']/crime_sum['폭력 발생']*100 crime_sum[['살인 검거', '살인 발생', '살인검거율']] del crime_sum['강간 검거'] del crime_sum['강도 검거'] del crime_sum['살인 검거'] del crime_sum['절도 검거'] del crime_sum['폭력 검거'] crime_sum[['살인 발생', '살인검거율']].head() 살인 발생 살인검거율 구별 강남구 13 76.923077 강동구 4 75.000000 강북구 7 114.285714 강서구 7 114.285714 관악구 9 88.888889

crime_sum.loc[crime_sum['강간검거율'] > 100, '강간검거율'] = 100 col_list = ['강간검거율','강도검거율','살인검거율','절도검거율','폭력검거율'] for column in col_list : crime_sum.loc[crime_sum[column] > 100, column] = 100 crime_sum.head() crime_sum.rename(columns = {'강간 발생' : '강간', '강도 발생' : '강도', '살인 발생' : '살인', '절도 발생' : '절도', '폭력 발생' : '폭력'}, inplace= True) crime_sum 강간 강도 살인 절도 폭력 강간검거율 강도검거율 살인검거율 절도검거율 폭력검거율 구별 강남구 449 21 13 3850 4284 77.728285 85.714286 76.923077 42.857143 86.484594 강동구 156 6 4 2366 2712 78.846154 100.000000 75.000000 33.347422 82.890855 강북구 153 14 7 1434 2649 82.352941 92.857143 100.000000 43.096234 88.637222 강서구 262 13 7 2096 3207 72.900763 100.000000 100.000000 60.114504 84.752105 관악구 320 12 9 2706 3298 69.062500 100.000000 88.888889 30.561715 80.109157 광진구 240 14 4 3026 2625 91.666667 100.000000 100.000000 42.200925 83.047619 구로구 281 15 8 2335 3007 58.362989 73.333333 75.000000 38.072805 80.877951 금천구 151 6 3 1567 2054 80.794702 100.000000 100.000000 56.668794 86.465433 노원구 197 7 10 2193 2723 61.421320 100.000000 100.000000 36.525308 85.530665 도봉구 102 9 3 1063 1487 100.000000 100.000000 100.000000 44.967074 87.626093 동대문구 173 13 5 1981 2548 84.393064 100.000000 100.000000 41.090358 87.401884 동작구 285 9 5 1865 1910 48.771930 55.555556 100.000000 35.442359 83.089005 마포구 294 14 8 2555 2983 84.013605 71.428571 100.000000 31.819961 84.445189 서대문구 154 5 2 1812 2056 80.519481 80.000000 100.000000 40.728477 83.219844 서초구 393 9 8 2635 2399 63.358779 66.666667 75.000000 41.404175 87.453105 성동구 126 9 4 1607 1612 94.444444 88.888889 100.000000 37.149969 86.538462 성북구 150 5 5 1785 2209 82.666667 80.000000 100.000000 41.512605 83.974649 송파구 220 13 11 3239 3295 80.909091 76.923077 90.909091 34.856437 84.552352 양천구 120 6 3 1890 2509 87.500000 50.000000 100.000000 35.555556 80.908729 영등포구 295 22 14 2964 3572 62.033898 90.909091 85.714286 32.995951 82.894737 용산구 194 14 5 1557 2050 89.175258 100.000000 100.000000 37.700706 83.121951 은평구 166 9 3 1914 2653 84.939759 66.666667 100.000000 37.147335 86.920467 종로구 211 11 6 2184 2293 76.303318 81.818182 83.333333 38.324176 84.212822 중구 170 9 3 2548 2224 65.294118 66.666667 66.666667 33.712716 88.309353 중랑구 187 11 13 2135 2847 79.144385 81.818182 92.307692 38.829040 84.545135

from sklearn import preprocessing as pp col = ['강간', '강도', '살인', '절도', '폭력'] x = crime_sum[col].values min_max_scaler = pp.MinMaxScaler() # 0 ~ 1 최대최소로 정규화 // 값이 범위가 너무 다르면 일치시키기 위해 x_scaled = min_max_scaler.fit_transform(x.astype(float)) crime_norm = pd.DataFrame(x_scaled, columns = col, index = crime_sum.index) crime_norm 강간 강도 살인 절도 폭력 구별 강남구 1.000000 0.941176 0.916667 1.000000 1.000000 강동구 0.155620 0.058824 0.166667 0.467528 0.437969 강북구 0.146974 0.529412 0.416667 0.133118 0.415445 강서구 0.461095 0.470588 0.416667 0.370649 0.614945 관악구 0.628242 0.411765 0.583333 0.589523 0.647479 광진구 0.397695 0.529412 0.166667 0.704342 0.406864 구로구 0.515850 0.588235 0.500000 0.456405 0.543439 금천구 0.141210 0.058824 0.083333 0.180840 0.202717 노원구 0.273775 0.117647 0.666667 0.405454 0.441902 도봉구 0.000000 0.235294 0.083333 0.000000 0.000000 동대문구 0.204611 0.470588 0.250000 0.329386 0.379335 동작구 0.527378 0.235294 0.250000 0.287765 0.151233 마포구 0.553314 0.529412 0.500000 0.535343 0.534859 서대문구 0.149856 0.000000 0.000000 0.268748 0.203432 서초구 0.838617 0.235294 0.500000 0.564047 0.326064 성동구 0.069164 0.235294 0.166667 0.195192 0.044691 성북구 0.138329 0.000000 0.250000 0.259060 0.258134 송파구 0.340058 0.470588 0.750000 0.780768 0.646407 양천구 0.051873 0.058824 0.083333 0.296735 0.365391 영등포구 0.556196 1.000000 1.000000 0.682095 0.745442 용산구 0.265130 0.529412 0.250000 0.177252 0.201287 은평구 0.184438 0.235294 0.083333 0.305346 0.416875 종로구 0.314121 0.352941 0.333333 0.402225 0.288166 중구 0.195965 0.235294 0.083333 0.532831 0.263497 중랑구 0.244957 0.352941 0.916667 0.384643 0.486235

저작자표시 비영리

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

23. titanic 분류 예측 \| KNN, SVM (0)	2021.11.24
22. auto-mpg \|\| 회귀분석 (0)	2021.11.24
20. 서울시 인구분석 \|\| 다중회귀 (0)	2021.11.23
19. 세계음주데이터2 (0)	2021.11.23
18. 세계음주 데이터 분석 (0)	2021.11.03

20. 서울시 인구분석 || 다중회귀

2021. 11. 23. 20:58

728x90

01. CCTV_in_Seoul

0.00MB

import pandas as pd cctv_seoul = pd.read_csv('01. CCTV_in_Seoul.csv', encoding = 'utf-8') cctv_seoul.rename(columns ={'기관명':'구별'}, inplace = True) cctv_seoul 구별 소계 2013년도 이전 2014년 2015년 2016년 0 강남구 2780 1292 430 584 932 1 강동구 773 379 99 155 377 2 강북구 748 369 120 138 204 3 강서구 884 388 258 184 81 4 관악구 1496 846 260 390 613 5 광진구 707 573 78 53 174 6 구로구 1561 1142 173 246 323 7 금천구 1015 674 51 269 354 8 노원구 1265 542 57 451 516 9 도봉구 485 238 159 42 386 10 동대문구 1294 1070 23 198 579 11 동작구 1091 544 341 103 314 12 마포구 574 314 118 169 379 13 서대문구 962 844 50 68 292 14 서초구 1930 1406 157 336 398 15 성동구 1062 730 91 241 265 16 성북구 1464 1009 78 360 204 17 송파구 618 529 21 68 463 18 양천구 2034 1843 142 30 467 19 영등포구 904 495 214 195 373 20 용산구 1624 1368 218 112 398 21 은평구 1873 1138 224 278 468 22 종로구 1002 464 314 211 630 23 중구 671 413 190 72 348 24 중랑구 660 509 121 177 109

pop_seoul = pd.read_excel('01. population_in_Seoul.xls', header = 2, usecols = 'B, D, G, J, N') pop_seoul.rename(columns = {pop_seoul.columns[0]:'구별', pop_seoul.columns[1]:'인구수', pop_seoul.columns[2]:'한국인', pop_seoul.columns[3]:'외국인', pop_seoul.columns[4]:'고령자'}, inplace = True) pop_seoul.head() 구별 인구수 한국인 외국인 고령자 0 합계 10197604.0 9926968.0 270636.0 1321458.0 1 종로구 162820.0 153589.0 9231.0 25425.0 2 중구 133240.0 124312.0 8928.0 20764.0 3 용산구 244203.0 229456.0 14747.0 36231.0 4 성동구 311244.0 303380.0 7864.0 39997.0

pop_seoul.info() <class 'pandas.core.frame.DataFrame'> RangeIndex: 27 entries, 0 to 26 Data columns (total 5 columns): # Column Non-Null Count Dtype --- ------ -------------- ----- 0 구별 26 non-null object 1 인구수 26 non-null float64 2 한국인 26 non-null float64 3 외국인 26 non-null float64 4 고령자 26 non-null float64 dtypes: float64(4), object(1) memory usage: 1.2+ KB

cctv_seoul['최근증가율'] = (cctv_seoul['2014년'] + cctv_seoul['2015년'] + cctv_seoul['2016년']) / \ cctv_seoul['2013년도 이전'] * 100 cctv_seoul.sort_values(by='최근증가율', ascending = False).head() 구별 소계 2013년도 이전 2014년 2015년 2016년 최근증가율 22 종로구 1002 464 314 211 630 248.922414 9 도봉구 485 238 159 42 386 246.638655 12 마포구 574 314 118 169 379 212.101911 8 노원구 1265 542 57 451 516 188.929889 1 강동구 773 379 99 155 377 166.490765

pop_seoul['외국인비율'] = pop_seoul['외국인'] /pop_seoul['인구수'] pop_seoul['고령자비율'] = pop_seoul['고령자'] /pop_seoul['인구수'] pop_seoul.head() 구별 인구수 한국인 외국인 고령자 외국인비율 고령자비율 0 합계 10197604.0 9926968.0 270636.0 1321458.0 0.026539 0.129585 1 종로구 162820.0 153589.0 9231.0 25425.0 0.056695 0.156154 2 중구 133240.0 124312.0 8928.0 20764.0 0.067007 0.155839 3 용산구 244203.0 229456.0 14747.0 36231.0 0.060388 0.148364 4 성동구 311244.0 303380.0 7864.0 39997.0 0.025266 0.128507

# 외국인비율 상위 5개 pop_seoul.sort_values(by = '외국인비율', ascending = False).head() 구별 인구수 한국인 외국인 고령자 외국인비율 고령자비율 19 영등포구 402985.0 368072.0 34913.0 52413.0 0.086636 0.130062 18 금천구 255082.0 236353.0 18729.0 32970.0 0.073423 0.129253 17 구로구 447874.0 416487.0 31387.0 56833.0 0.070080 0.126895 2 중구 133240.0 124312.0 8928.0 20764.0 0.067007 0.155839 3 용산구 244203.0 229456.0 14747.0 36231.0 0.060388 0.148364

data_result = pd.merge(cctv_seoul, pop_seoul, on = '구별') data_result.head() 구별 소계 2013년도 이전 2014년 2015년 2016년 최근증가율 인구수 한국인 외국인 고령자 외국인비율 고령자비율 0 강남구 2780 1292 430 584 932 150.619195 570500.0 565550.0 4950.0 63167.0 0.008677 0.110722 1 강동구 773 379 99 155 377 166.490765 453233.0 449019.0 4214.0 54622.0 0.009298 0.120516 2 강북구 748 369 120 138 204 125.203252 330192.0 326686.0 3506.0 54813.0 0.010618 0.166003 3 강서구 884 388 258 184 81 134.793814 603772.0 597248.0 6524.0 72548.0 0.010805 0.120158 4 관악구 1496 846 260 390 613 149.290780 525515.0 507203.0 18312.0 68082.0 0.034846 0.129553

del data_result['2013년도 이전'] del data_result['2014년'] del data_result['2015년'] del data_result['2016년'] data_result.head() 구별 소계 최근증가율 인구수 한국인 외국인 고령자 외국인비율 고령자비율 0 강남구 2780 150.619195 570500.0 565550.0 4950.0 63167.0 0.008677 0.110722 1 강동구 773 166.490765 453233.0 449019.0 4214.0 54622.0 0.009298 0.120516 2 강북구 748 125.203252 330192.0 326686.0 3506.0 54813.0 0.010618 0.166003 3 강서구 884 134.793814 603772.0 597248.0 6524.0 72548.0 0.010805 0.120158 4 관악구 1496 149.290780 525515.0 507203.0 18312.0 68082.0 0.034846 0.129553

data_result.set_index('구별', inplace = True) data_result.head() 소계 최근증가율 인구수 한국인 외국인 고령자 외국인비율 고령자비율 구별 강남구 2780 150.619195 570500.0 565550.0 4950.0 63167.0 0.008677 0.110722 강동구 773 166.490765 453233.0 449019.0 4214.0 54622.0 0.009298 0.120516 강북구 748 125.203252 330192.0 326686.0 3506.0 54813.0 0.010618 0.166003 강서구 884 134.793814 603772.0 597248.0 6524.0 72548.0 0.010805 0.120158 관악구 1496 149.290780 525515.0 507203.0 18312.0 68082.0 0.034846 0.129553

data_result.corr(method='pearson') 소계 최근증가율 인구수 한국인 외국인 고령자 외국인비율 고령자비율 소계 1.000000 -0.343016 0.306342 0.304287 -0.023786 0.255196 -0.136074 -0.280786 최근증가율 -0.343016 1.000000 -0.093068 -0.082511 -0.150463 -0.070969 -0.044042 0.185089 인구수 0.306342 -0.093068 1.000000 0.998061 -0.153371 0.932667 -0.591939 -0.669462 한국인 0.304287 -0.082511 0.998061 1.000000 -0.214576 0.931636 -0.637911 -0.660812 외국인 -0.023786 -0.150463 -0.153371 -0.214576 1.000000 -0.155381 0.838904 -0.014055 고령자 0.255196 -0.070969 0.932667 0.931636 -0.155381 1.000000 -0.606088 -0.380468 외국인비율 -0.136074 -0.044042 -0.591939 -0.637911 0.838904 -0.606088 1.000000 0.267348 고령자비율 -0.280786 0.185089 -0.669462 -0.660812 -0.014055 -0.380468 0.267348 1.000000

import numpy as np np.corrcoef(data_result['고령자비율'], data_result['소계']) array([[ 1. , -0.28078554], [-0.28078554, 1. ]]) np.corrcoef(data_result['외국인비율'], data_result['소계']) array([[ 1. , -0.13607433], [-0.13607433, 1. ]]) np.corrcoef(data_result['인구수'], data_result['소계']) array([[1. , 0.30634228], [0.30634228, 1. ]])

#matplot import matplotlib.pyplot as plt from matplotlib import font_manager, rc plt.figure() plt.rc('font', family = 'Malgun Gothic') data_result['소계'].sort_values(ascending = True).plot(kind='barh', grid=True, figsize = (10,10)) plt.show()

# 인구대비 cctv 비율컬럼 data_result['cctv비율'] = data_result['소계'] / data_result['인구수'] * 100 data_result.head() 소계 최근증가율 인구수 한국인 외국인 고령자 외국인비율 고령자비율 cctv비율 구별 강남구 2780 150.619195 570500.0 565550.0 4950.0 63167.0 0.008677 0.110722 0.487292 강동구 773 166.490765 453233.0 449019.0 4214.0 54622.0 0.009298 0.120516 0.170552 강북구 748 125.203252 330192.0 326686.0 3506.0 54813.0 0.010618 0.166003 0.226535 강서구 884 134.793814 603772.0 597248.0 6524.0 72548.0 0.010805 0.120158 0.146413 관악구 1496 149.290780 525515.0 507203.0 18312.0 68082.0 0.034846 0.129553 0.284673

data_result['cctv비율'].sort_values().plot(kind='barh', grid = True, figsize = (10,10)) plt.show()

# 산점도 plt.figure(figsize = (6,6)) plt.scatter(data_result['인구수'], data_result['소계'], s= 50) plt.xlabel('인구수') plt.ylabel('cctv갯수') plt.grid() plt.show()

# 인구수와 소계 산점도, 회귀선 작성 # polyfit 최소제곱법을 이용한 상수값, 1 : 차수 fpl = np.polyfit(data_result['인구수'], data_result['소계'], 1) # ㅣ 직선 f1 = np.poly1d(fpl) fx = np.linspace(100000, 700000, 100) plt.figure(figsize = (10, 10)) plt.scatter(data_result['인구수'], data_result['소계'], s=50) plt.plot(fx, f1(fx), ls='dashed', lw=3, color='g') plt.xlabel('인구수') plt.ylabel('cctv') plt.grid() plt.show()

# 인구수와 소계 산점도, 회귀선 작성 # polyfit 최소제곱법을 이용한 상수값, 1 : 차수 fpl = np.polyfit(data_result['인구수'], data_result['소계'], 4) # ㅣ 직선 f1 = np.poly1d(fpl) # 인구수에 맞는 y값 fx = np.linspace(100000, 700000, 100) plt.figure(figsize = (10, 10)) plt.scatter(data_result['인구수'], data_result['소계'], s=50) plt.plot(fx, f1(fx), ls='dashed', lw=3, color='g') plt.xlabel('인구수') plt.ylabel('cctv') plt.grid() plt.show()

import numpy as np x = np.array([0., 1., 2., 3., 4., 5., 6., 7., 8., 9., 10.]) y = np.array([4.23620563, 6.18696492, 2.83930821, 5.00923197, 11.51299327, 12.91581993, 14.51838241, 14.34881875, 18.13566499, 20.1408104, 21.9872241]) fit1 = np.polyfit(x, y, 1) # 2개 상수 (+ 절편) fit2 = np.polyfit(x, y, 2) # 3개 상수 (+ 절편) fit3 = np.polyfit(x, y, 3) # 4개 상수 (+ 절편) print(fit1) print(fit2) print(fit3) # [1.92858279 2.34176099] # [0.05915413 1.33704154 3.22907288] # [-0.02808825 0.48047788 -0.26960637 4.24024989]

num = len(x) for i in range(num) : fit1 = 1.92858279*x + 2.34176099 fit2 = 0.05915413*x**2 + 1.33704154*x + 3.22907288 fit3 = - 0.02808825*x**3 + 0.48047788*x**2 - 0.26960637*x + 4.24024989 print(fit3) # [ 4.24024989 4.42303315 5.39824267 6.99734895 9.05182249 11.39313379 # 13.85275335 16.26215167 18.45279925 20.25616659 21.50372419]

# xy 산점도 와 회귀선 plt.scatter(x, y) plt.plot(x, y) plt.plot(x, fit1) plt.show()

# xy 산점도 와 회귀선 plt.scatter(x, y) plt.plot(x, y) plt.plot(x, fit3) plt.show()

# 산점도 + 회귀선, 산점도에 색상을 회귀선과의 거리로 표시 # 회귀선을 위한 상수 fpl = np.polyfit(data_result['인구수'], data_result['소계'], 2) # fpl상수값을 이용하여 y값을 계산하기 위한 함수 f1 = np.poly1d(fpl) # X축값, 10만 ~ 70만 까지 100등분 fx = np.linspace(100000, 700000, 100) # data_result 인구수에 맞는 회귀선의 y값 # 절대값 data_result['오차'] = np.abs(data_result['소계'] - f1(data_result['인구수'])) df_sort = data_result.sort_values(by = '오차', ascending = False) # 그래프 작성 plt.figure(figsize = (14,10)) plt.scatter(data_result['인구수'], data_result['소계'], c =data_result['오차'], s=50) plt.plot(fx, f1(fx), ls = 'dashed', lw = 3, color = 'g') # 점에 구 이름 표시 for n in range(10) : # 라벨링 // 절대값 오차가 많은 구10개 정보 표시 plt.text(df_sort['인구수'][n]*1.02, df_sort['소계'][n]*0.98, # 약간 밑으로 df_sort.index[n], fontsize = 15) plt.xlabel('인구수') plt.ylabel('cctv갯수') plt.colorbar() plt.grid() plt.show()

저작자표시 비영리

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

22. auto-mpg \|\| 회귀분석 (0)	2021.11.24
21. 서울시 범죄율 분석 \|\| MinMaxscalimg (0)	2021.11.24
19. 세계음주데이터2 (0)	2021.11.23
18. 세계음주 데이터 분석 (0)	2021.11.03
17. 서울 기온 분석 (0)	2021.11.02

19. 세계음주데이터2

2021. 11. 23. 20:04

728x90

drinks.csv

0.00MB

from scipy import stats
import pandas as pd
drinks = pd.read_csv('drinks.csv')
drinks['continent'] = drinks['continent'].fillna('OT')
drinks.info

<bound method DataFrame.info of          country  beer_servings  spirit_servings  wine_servings  \
0    Afghanistan              0                0              0   
1        Albania             89              132             54   
2        Algeria             25                0             14   
3        Andorra            245              138            312   
4         Angola            217               57             45   
..           ...            ...              ...            ...   
188    Venezuela            333              100              3   
189      Vietnam            111                2              1   
190        Yemen              6                0              0   
191       Zambia             32               19              4   
192     Zimbabwe             64               18              4   

     total_litres_of_pure_alcohol continent  
0                             0.0        AS  
1                             4.9        EU  
2                             0.7        AF  
3                            12.4        EU  
4                             5.9        AF  
..                            ...       ...  
188                           7.7        SA  
189                           2.0        AS  
190                           0.1        AS  
191                           2.5        AF  
192                           4.7        AF  

[193 rows x 6 columns]>

africa = drinks.loc[drinks['continent']=='AF']
europe = drinks.loc[drinks['continent']=='EU']
# 두집단간 평균의 차이
tTestResult = stats.ttest_ind(africa['beer_servings'], europe['beer_servings'])
tTestResultDiffVar = stats.ttest_ind(africa['beer_servings'], europe['beer_servings'], equal_var = False)

# 두집단의 분산이 같다 가설
print(tTestResult)

# Ttest_indResult(statistic=-7.267986335644365, pvalue=9.719556422442453e-11)

# 두집단의 분산이 다르다 가설
print(tTestResultDiffVar)

# Ttest_indResult(statistic=-7.143520192189803, pvalue=2.9837787864303205e-10)

- t-statistic : 평균차이, 음수 : 뒤쪽 데이터의 평균 큰 경우, 검정 통계
- p-value : 유의확률, 결과가 0, 두집단의 평균이 같지 않다. => 귀무가설이 기각, 맞다틀리다
- 귀무가설 : 현재가설이 맞지 않다를 증명 // 예상되는 가설
- 대립가설 : 귀무가설의 반대되는 가설,
- 아프리카와 유럽의 맥주소비량의 차이는 확률적으로 다르다
- => 통계적으로 유의미하다

# 대한민국은 얼마나 술을 독하게 마실까?
drinks['total_servings'] =  drinks['beer_servings'] + drinks['spirit_servings']+drinks['wine_servings']
drinks.head()

	country	beer_servings	spirit_servings	wine_servings	total_litres_of_pure_alcohol	continent	total_servings
0	Afghanistan	0	0	0	0.0	AS	0
1	Albania	89	132	54	4.9	EU	275
2	Algeria	25	0	14	0.7	AF	39
3	Andorra	245	138	312	12.4	EU	695
4	Angola	217	57	45	5.9	AF	319

drinks['alcohol_rate'] = drinks['total_litres_of_pure_alcohol'] / drinks['total_servings']
# alcohol rate , 분모가 0이면 결측값이 생김
drinks.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 193 entries, 0 to 192
Data columns (total 8 columns):
 #   Column                        Non-Null Count  Dtype  
---  ------                        --------------  -----  
 0   country                       193 non-null    object 
 1   beer_servings                 193 non-null    int64  
 2   spirit_servings               193 non-null    int64  
 3   wine_servings                 193 non-null    int64  
 4   total_litres_of_pure_alcohol  193 non-null    float64
 5   continent                     193 non-null    object 
 6   total_servings                193 non-null    int64  
 7   alcohol_rate                  180 non-null    float64
dtypes: float64(2), int64(4), object(2)
memory usage: 12.2+ KB

drinks['alcohol_rate'].fillna(0, inplace = True)
drinks.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 193 entries, 0 to 192
Data columns (total 8 columns):
 #   Column                        Non-Null Count  Dtype  
---  ------                        --------------  -----  
 0   country                       193 non-null    object 
 1   beer_servings                 193 non-null    int64  
 2   spirit_servings               193 non-null    int64  
 3   wine_servings                 193 non-null    int64  
 4   total_litres_of_pure_alcohol  193 non-null    float64
 5   continent                     193 non-null    object 
 6   total_servings                193 non-null    int64  
 7   alcohol_rate                  193 non-null    float64
dtypes: float64(2), int64(4), object(2)
memory usage: 12.2+ KB

country_alcohol_rank = drinks[['country', 'alcohol_rate']]
country_alcohol_rank = country_alcohol_rank.sort_values(by = ['alcohol_rate'], ascending = False)
country_alcohol_rank.head()

	country	alcohol_rate
63	Gambia	0.266667
153	Sierra Leone	0.223333
124	Nigeria	0.185714
179	Uganda	0.153704
142	Rwanda	0.151111

import numpy as np
import matplotlib.pyplot as plt
country_list = country_alcohol_rank.country.tolist()
x_pos = np.arange(len(country_list))
rank = country_alcohol_rank.alcohol_rate.tolist()
country_list.index("South Korea")

bar_list = plt.bar(x_pos, rank)
bar_list[country_list.index('South Korea')].set_color('r')
plt.ylabel('alcohol rate')
plt.title('liquor drink rank by country')
plt.axis([0, 200, 0, 0.3])

korea_rank = country_list.index('South Korea')
korea_alc_rate = country_alcohol_rank[country_alcohol_rank['country'] == 'South Korea']['alcohol_rate'].values[0]
plt.annotate('South korea :' + str(korea_rank + 1), xy = (korea_rank, korea_alc_rate), 
            xytext = (korea_rank + 10, korea_alc_rate + 0.05),
            arrowprops = dict(facecolor = 'red', shrink = 0.05))
plt.show()

#  전체 소비량을 막대그래프로 작성
country_serving_rank = drinks[['country','total_servings']]
country_serving_rank = country_serving_rank.sort_values(by=['total_servings'], ascending=0)
country_serving_rank.head()

	country	total_servings
3	Andorra	695
68	Grenada	665
45	Czech Republic	665
61	France	648
141	Russian Federation	646

# 그래프 작성하기 
country_list = country_serving_rank.country.tolist()
x_pos = np.arange(len(country_list))
rank = country_serving_rank.total_servings.tolist()

bar_list = plt.bar(x_pos, rank)
bar_list[country_list.index('South Korea')].set_color('r')
plt.ylabel('alcohol rate')
plt.title('liquor drink rank by country')
plt.axis([0, 200, 0, 700])

korea_rank = country_list.index('South Korea')
korea_serving_rate = country_serving_rank[country_serving_rank['country'] == 'South Korea']['total_servings'].values[0]
plt.annotate('South korea :' + str(korea_rank + 1), xy = (korea_rank, korea_serving_rate), 
            xytext = (korea_rank + 10, korea_serving_rate + 0.05),
            arrowprops = dict(facecolor = 'red', shrink = 0.05))
plt.show()

저작자표시 비영리

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

21. 서울시 범죄율 분석 \|\| MinMaxscalimg (0)	2021.11.24
20. 서울시 인구분석 \|\| 다중회귀 (0)	2021.11.23
18. 세계음주 데이터 분석 (0)	2021.11.03
17. 서울 기온 분석 (0)	2021.11.02
16. EDA, 멕시코식당 주문 CHIPOTLE (0)	2021.10.28

제2회 성남여성 창업아이디어 경진대회 수상

2021. 11. 17. 17:38

728x90

재활용품 분류 사물인식 딥러닝모델 아이디어로 한국 지역난방공사상 수상

저작자표시 비영리

39일차 - 논문 수업 (GAN)

2021. 11. 16. 19:36

728x90

Image-to-Image Translation with Conditional Adversarial Networks

Conditional GAN

GAN에서 조건을 부여하면 이미지를 생성시키는 모델
이미지를 변환할 때 생성 모델과 판별 모델에 조건으로 이미지를 넣어 학습하면
생성 모델은 입력 이미지에 따르는 결과 이미지를 생성할 수 있게 된다

흑백 이미지를 컬러와 하거나, 윤곽이 주어지면 물체를 그리는 문제와 같은 것들을 해결 할 수 있다

픽셀 하나 하나를 바꾼다

Generator

Encoder와 decoder가 skip connection(combination)으로 연결되어 있는 U-net을 사용한다
이미지의 맥락을 보존시키는데 효과적인 구조이다

Discriminator

기존의 discriminator는 이미지 전체에 대해 fake/real 판단을 내리는 반면
PatchGAN은 이미지를 작은 patch들로 쪼개 그 patch에 대해 fake/real를 판단한다
이렇게 함으로써 L1 loss가 만들어낸 전체적인 특징과 더불어 좀 더 디테일한 특징들을 이미지에서 살려낼 수 있다

부가적 이점으로는 전체 이미지보다 작은 크기에 적용하기 때문에 파라미터 수가 적고 좀 더 빠른 속도로 연산이 가능하다
또한 더 큰 사이즈의 이미지에 대해 일반화 하여 적용시킬 수 있는 장점이 있다

PatchGAN discriminator

원본 이미지에 70x70만큼 움직이면서 window연산을 하고 연산 결과 30x30 크기 이미지가 나온다
이때 30x30 크기 이미지를 통해 각각의 부분을 예측한다

convolution은 특징이 있는지 없는지 여부를 찾고 주변 데이터와의 상관관계를 따지지 않기 때문에
전체 이미지를 통해 예측을 하게 되면 정확한 예측이 불가능하다
따라서 부분 부분의 이미지를 통해 예측을 하는 방식을 사용해야 한다

import tensorflow as tf 

resnet = tf.keras.applications.ResNet152V2()

resnet.summary() # Add를 사용해서 결과를 유지했다

Model: "resnet152v2"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_2 (InputLayer)            [(None, 224, 224, 3) 0                                            
__________________________________________________________________________________________________
conv1_pad (ZeroPadding2D)       (None, 230, 230, 3)  0           input_2[0][0]                    
__________________________________________________________________________________________________
conv1_conv (Conv2D)             (None, 112, 112, 64) 9472        conv1_pad[0][0]                  
__________________________________________________________________________________________________
pool1_pad (ZeroPadding2D)       (None, 114, 114, 64) 0           conv1_conv[0][0]                 
__________________________________________________________________________________________________
pool1_pool (MaxPooling2D)       (None, 56, 56, 64)   0           pool1_pad[0][0]                  
__________________________________________________________________________________________________
conv2_block1_preact_bn (BatchNo (None, 56, 56, 64)   256         pool1_pool[0][0]                 
__________________________________________________________________________________________________
conv2_block1_preact_relu (Activ (None, 56, 56, 64)   0           conv2_block1_preact_bn[0][0]     
__________________________________________________________________________________________________
conv2_block1_1_conv (Conv2D)    (None, 56, 56, 64)   4096        conv2_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv2_block1_1_bn (BatchNormali (None, 56, 56, 64)   256         conv2_block1_1_conv[0][0]        
__________________________________________________________________________________________________
conv2_block1_1_relu (Activation (None, 56, 56, 64)   0           conv2_block1_1_bn[0][0]          
__________________________________________________________________________________________________
conv2_block1_2_pad (ZeroPadding (None, 58, 58, 64)   0           conv2_block1_1_relu[0][0]        
__________________________________________________________________________________________________
conv2_block1_2_conv (Conv2D)    (None, 56, 56, 64)   36864       conv2_block1_2_pad[0][0]         
__________________________________________________________________________________________________
conv2_block1_2_bn (BatchNormali (None, 56, 56, 64)   256         conv2_block1_2_conv[0][0]        
__________________________________________________________________________________________________
conv2_block1_2_relu (Activation (None, 56, 56, 64)   0           conv2_block1_2_bn[0][0]          
__________________________________________________________________________________________________
conv2_block1_0_conv (Conv2D)    (None, 56, 56, 256)  16640       conv2_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv2_block1_3_conv (Conv2D)    (None, 56, 56, 256)  16640       conv2_block1_2_relu[0][0]        
__________________________________________________________________________________________________
conv2_block1_out (Add)          (None, 56, 56, 256)  0           conv2_block1_0_conv[0][0]        
                                                                 conv2_block1_3_conv[0][0]        
__________________________________________________________________________________________________
conv2_block2_preact_bn (BatchNo (None, 56, 56, 256)  1024        conv2_block1_out[0][0]           
__________________________________________________________________________________________________
conv2_block2_preact_relu (Activ (None, 56, 56, 256)  0           conv2_block2_preact_bn[0][0]     
__________________________________________________________________________________________________
conv2_block2_1_conv (Conv2D)    (None, 56, 56, 64)   16384       conv2_block2_preact_relu[0][0]   
__________________________________________________________________________________________________
conv2_block2_1_bn (BatchNormali (None, 56, 56, 64)   256         conv2_block2_1_conv[0][0]        
__________________________________________________________________________________________________
conv2_block2_1_relu (Activation (None, 56, 56, 64)   0           conv2_block2_1_bn[0][0]          
__________________________________________________________________________________________________
conv2_block2_2_pad (ZeroPadding (None, 58, 58, 64)   0           conv2_block2_1_relu[0][0]        
__________________________________________________________________________________________________
conv2_block2_2_conv (Conv2D)    (None, 56, 56, 64)   36864       conv2_block2_2_pad[0][0]         
__________________________________________________________________________________________________
conv2_block2_2_bn (BatchNormali (None, 56, 56, 64)   256         conv2_block2_2_conv[0][0]        
__________________________________________________________________________________________________
conv2_block2_2_relu (Activation (None, 56, 56, 64)   0           conv2_block2_2_bn[0][0]          
__________________________________________________________________________________________________
conv2_block2_3_conv (Conv2D)    (None, 56, 56, 256)  16640       conv2_block2_2_relu[0][0]        
__________________________________________________________________________________________________
conv2_block2_out (Add)          (None, 56, 56, 256)  0           conv2_block1_out[0][0]           
                                                                 conv2_block2_3_conv[0][0]        
__________________________________________________________________________________________________
conv2_block3_preact_bn (BatchNo (None, 56, 56, 256)  1024        conv2_block2_out[0][0]           
__________________________________________________________________________________________________
conv2_block3_preact_relu (Activ (None, 56, 56, 256)  0           conv2_block3_preact_bn[0][0]     
__________________________________________________________________________________________________
conv2_block3_1_conv (Conv2D)    (None, 56, 56, 64)   16384       conv2_block3_preact_relu[0][0]   
__________________________________________________________________________________________________
conv2_block3_1_bn (BatchNormali (None, 56, 56, 64)   256         conv2_block3_1_conv[0][0]        
__________________________________________________________________________________________________
conv2_block3_1_relu (Activation (None, 56, 56, 64)   0           conv2_block3_1_bn[0][0]          
__________________________________________________________________________________________________
conv2_block3_2_pad (ZeroPadding (None, 58, 58, 64)   0           conv2_block3_1_relu[0][0]        
__________________________________________________________________________________________________
conv2_block3_2_conv (Conv2D)    (None, 28, 28, 64)   36864       conv2_block3_2_pad[0][0]         
__________________________________________________________________________________________________
conv2_block3_2_bn (BatchNormali (None, 28, 28, 64)   256         conv2_block3_2_conv[0][0]        
__________________________________________________________________________________________________
conv2_block3_2_relu (Activation (None, 28, 28, 64)   0           conv2_block3_2_bn[0][0]          
__________________________________________________________________________________________________
max_pooling2d_3 (MaxPooling2D)  (None, 28, 28, 256)  0           conv2_block2_out[0][0]           
__________________________________________________________________________________________________
conv2_block3_3_conv (Conv2D)    (None, 28, 28, 256)  16640       conv2_block3_2_relu[0][0]        
__________________________________________________________________________________________________
conv2_block3_out (Add)          (None, 28, 28, 256)  0           max_pooling2d_3[0][0]            
                                                                 conv2_block3_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block1_preact_bn (BatchNo (None, 28, 28, 256)  1024        conv2_block3_out[0][0]           
__________________________________________________________________________________________________
conv3_block1_preact_relu (Activ (None, 28, 28, 256)  0           conv3_block1_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block1_1_conv (Conv2D)    (None, 28, 28, 128)  32768       conv3_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block1_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block1_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block1_1_relu (Activation (None, 28, 28, 128)  0           conv3_block1_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block1_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block1_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block1_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block1_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block1_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block1_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block1_2_relu (Activation (None, 28, 28, 128)  0           conv3_block1_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block1_0_conv (Conv2D)    (None, 28, 28, 512)  131584      conv3_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block1_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block1_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block1_out (Add)          (None, 28, 28, 512)  0           conv3_block1_0_conv[0][0]        
                                                                 conv3_block1_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block2_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block1_out[0][0]           
__________________________________________________________________________________________________
conv3_block2_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block2_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block2_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block2_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block2_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block2_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block2_1_relu (Activation (None, 28, 28, 128)  0           conv3_block2_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block2_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block2_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block2_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block2_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block2_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block2_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block2_2_relu (Activation (None, 28, 28, 128)  0           conv3_block2_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block2_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block2_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block2_out (Add)          (None, 28, 28, 512)  0           conv3_block1_out[0][0]           
                                                                 conv3_block2_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block3_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block2_out[0][0]           
__________________________________________________________________________________________________
conv3_block3_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block3_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block3_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block3_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block3_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block3_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block3_1_relu (Activation (None, 28, 28, 128)  0           conv3_block3_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block3_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block3_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block3_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block3_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block3_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block3_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block3_2_relu (Activation (None, 28, 28, 128)  0           conv3_block3_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block3_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block3_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block3_out (Add)          (None, 28, 28, 512)  0           conv3_block2_out[0][0]           
                                                                 conv3_block3_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block4_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block3_out[0][0]           
__________________________________________________________________________________________________
conv3_block4_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block4_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block4_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block4_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block4_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block4_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block4_1_relu (Activation (None, 28, 28, 128)  0           conv3_block4_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block4_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block4_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block4_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block4_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block4_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block4_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block4_2_relu (Activation (None, 28, 28, 128)  0           conv3_block4_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block4_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block4_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block4_out (Add)          (None, 28, 28, 512)  0           conv3_block3_out[0][0]           
                                                                 conv3_block4_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block5_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block4_out[0][0]           
__________________________________________________________________________________________________
conv3_block5_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block5_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block5_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block5_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block5_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block5_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block5_1_relu (Activation (None, 28, 28, 128)  0           conv3_block5_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block5_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block5_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block5_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block5_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block5_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block5_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block5_2_relu (Activation (None, 28, 28, 128)  0           conv3_block5_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block5_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block5_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block5_out (Add)          (None, 28, 28, 512)  0           conv3_block4_out[0][0]           
                                                                 conv3_block5_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block6_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block5_out[0][0]           
__________________________________________________________________________________________________
conv3_block6_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block6_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block6_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block6_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block6_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block6_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block6_1_relu (Activation (None, 28, 28, 128)  0           conv3_block6_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block6_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block6_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block6_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block6_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block6_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block6_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block6_2_relu (Activation (None, 28, 28, 128)  0           conv3_block6_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block6_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block6_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block6_out (Add)          (None, 28, 28, 512)  0           conv3_block5_out[0][0]           
                                                                 conv3_block6_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block7_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block6_out[0][0]           
__________________________________________________________________________________________________
conv3_block7_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block7_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block7_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block7_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block7_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block7_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block7_1_relu (Activation (None, 28, 28, 128)  0           conv3_block7_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block7_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block7_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block7_2_conv (Conv2D)    (None, 28, 28, 128)  147456      conv3_block7_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block7_2_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block7_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block7_2_relu (Activation (None, 28, 28, 128)  0           conv3_block7_2_bn[0][0]          
__________________________________________________________________________________________________
conv3_block7_3_conv (Conv2D)    (None, 28, 28, 512)  66048       conv3_block7_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block7_out (Add)          (None, 28, 28, 512)  0           conv3_block6_out[0][0]           
                                                                 conv3_block7_3_conv[0][0]        
__________________________________________________________________________________________________
conv3_block8_preact_bn (BatchNo (None, 28, 28, 512)  2048        conv3_block7_out[0][0]           
__________________________________________________________________________________________________
conv3_block8_preact_relu (Activ (None, 28, 28, 512)  0           conv3_block8_preact_bn[0][0]     
__________________________________________________________________________________________________
conv3_block8_1_conv (Conv2D)    (None, 28, 28, 128)  65536       conv3_block8_preact_relu[0][0]   
__________________________________________________________________________________________________
conv3_block8_1_bn (BatchNormali (None, 28, 28, 128)  512         conv3_block8_1_conv[0][0]        
__________________________________________________________________________________________________
conv3_block8_1_relu (Activation (None, 28, 28, 128)  0           conv3_block8_1_bn[0][0]          
__________________________________________________________________________________________________
conv3_block8_2_pad (ZeroPadding (None, 30, 30, 128)  0           conv3_block8_1_relu[0][0]        
__________________________________________________________________________________________________
conv3_block8_2_conv (Conv2D)    (None, 14, 14, 128)  147456      conv3_block8_2_pad[0][0]         
__________________________________________________________________________________________________
conv3_block8_2_bn (BatchNormali (None, 14, 14, 128)  512         conv3_block8_2_conv[0][0]        
__________________________________________________________________________________________________
conv3_block8_2_relu (Activation (None, 14, 14, 128)  0           conv3_block8_2_bn[0][0]          
__________________________________________________________________________________________________
max_pooling2d_4 (MaxPooling2D)  (None, 14, 14, 512)  0           conv3_block7_out[0][0]           
__________________________________________________________________________________________________
conv3_block8_3_conv (Conv2D)    (None, 14, 14, 512)  66048       conv3_block8_2_relu[0][0]        
__________________________________________________________________________________________________
conv3_block8_out (Add)          (None, 14, 14, 512)  0           max_pooling2d_4[0][0]            
                                                                 conv3_block8_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block1_preact_bn (BatchNo (None, 14, 14, 512)  2048        conv3_block8_out[0][0]           
__________________________________________________________________________________________________
conv4_block1_preact_relu (Activ (None, 14, 14, 512)  0           conv4_block1_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block1_1_conv (Conv2D)    (None, 14, 14, 256)  131072      conv4_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block1_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block1_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block1_1_relu (Activation (None, 14, 14, 256)  0           conv4_block1_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block1_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block1_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block1_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block1_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block1_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block1_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block1_2_relu (Activation (None, 14, 14, 256)  0           conv4_block1_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block1_0_conv (Conv2D)    (None, 14, 14, 1024) 525312      conv4_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block1_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block1_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block1_out (Add)          (None, 14, 14, 1024) 0           conv4_block1_0_conv[0][0]        
                                                                 conv4_block1_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block2_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block1_out[0][0]           
__________________________________________________________________________________________________
conv4_block2_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block2_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block2_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block2_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block2_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block2_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block2_1_relu (Activation (None, 14, 14, 256)  0           conv4_block2_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block2_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block2_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block2_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block2_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block2_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block2_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block2_2_relu (Activation (None, 14, 14, 256)  0           conv4_block2_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block2_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block2_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block2_out (Add)          (None, 14, 14, 1024) 0           conv4_block1_out[0][0]           
                                                                 conv4_block2_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block3_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block2_out[0][0]           
__________________________________________________________________________________________________
conv4_block3_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block3_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block3_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block3_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block3_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block3_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block3_1_relu (Activation (None, 14, 14, 256)  0           conv4_block3_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block3_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block3_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block3_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block3_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block3_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block3_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block3_2_relu (Activation (None, 14, 14, 256)  0           conv4_block3_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block3_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block3_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block3_out (Add)          (None, 14, 14, 1024) 0           conv4_block2_out[0][0]           
                                                                 conv4_block3_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block4_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block3_out[0][0]           
__________________________________________________________________________________________________
conv4_block4_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block4_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block4_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block4_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block4_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block4_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block4_1_relu (Activation (None, 14, 14, 256)  0           conv4_block4_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block4_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block4_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block4_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block4_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block4_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block4_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block4_2_relu (Activation (None, 14, 14, 256)  0           conv4_block4_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block4_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block4_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block4_out (Add)          (None, 14, 14, 1024) 0           conv4_block3_out[0][0]           
                                                                 conv4_block4_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block5_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block4_out[0][0]           
__________________________________________________________________________________________________
conv4_block5_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block5_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block5_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block5_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block5_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block5_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block5_1_relu (Activation (None, 14, 14, 256)  0           conv4_block5_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block5_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block5_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block5_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block5_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block5_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block5_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block5_2_relu (Activation (None, 14, 14, 256)  0           conv4_block5_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block5_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block5_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block5_out (Add)          (None, 14, 14, 1024) 0           conv4_block4_out[0][0]           
                                                                 conv4_block5_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block6_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block5_out[0][0]           
__________________________________________________________________________________________________
conv4_block6_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block6_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block6_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block6_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block6_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block6_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block6_1_relu (Activation (None, 14, 14, 256)  0           conv4_block6_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block6_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block6_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block6_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block6_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block6_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block6_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block6_2_relu (Activation (None, 14, 14, 256)  0           conv4_block6_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block6_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block6_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block6_out (Add)          (None, 14, 14, 1024) 0           conv4_block5_out[0][0]           
                                                                 conv4_block6_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block7_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block6_out[0][0]           
__________________________________________________________________________________________________
conv4_block7_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block7_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block7_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block7_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block7_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block7_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block7_1_relu (Activation (None, 14, 14, 256)  0           conv4_block7_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block7_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block7_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block7_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block7_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block7_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block7_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block7_2_relu (Activation (None, 14, 14, 256)  0           conv4_block7_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block7_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block7_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block7_out (Add)          (None, 14, 14, 1024) 0           conv4_block6_out[0][0]           
                                                                 conv4_block7_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block8_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block7_out[0][0]           
__________________________________________________________________________________________________
conv4_block8_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block8_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block8_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block8_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block8_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block8_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block8_1_relu (Activation (None, 14, 14, 256)  0           conv4_block8_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block8_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block8_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block8_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block8_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block8_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block8_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block8_2_relu (Activation (None, 14, 14, 256)  0           conv4_block8_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block8_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block8_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block8_out (Add)          (None, 14, 14, 1024) 0           conv4_block7_out[0][0]           
                                                                 conv4_block8_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block9_preact_bn (BatchNo (None, 14, 14, 1024) 4096        conv4_block8_out[0][0]           
__________________________________________________________________________________________________
conv4_block9_preact_relu (Activ (None, 14, 14, 1024) 0           conv4_block9_preact_bn[0][0]     
__________________________________________________________________________________________________
conv4_block9_1_conv (Conv2D)    (None, 14, 14, 256)  262144      conv4_block9_preact_relu[0][0]   
__________________________________________________________________________________________________
conv4_block9_1_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block9_1_conv[0][0]        
__________________________________________________________________________________________________
conv4_block9_1_relu (Activation (None, 14, 14, 256)  0           conv4_block9_1_bn[0][0]          
__________________________________________________________________________________________________
conv4_block9_2_pad (ZeroPadding (None, 16, 16, 256)  0           conv4_block9_1_relu[0][0]        
__________________________________________________________________________________________________
conv4_block9_2_conv (Conv2D)    (None, 14, 14, 256)  589824      conv4_block9_2_pad[0][0]         
__________________________________________________________________________________________________
conv4_block9_2_bn (BatchNormali (None, 14, 14, 256)  1024        conv4_block9_2_conv[0][0]        
__________________________________________________________________________________________________
conv4_block9_2_relu (Activation (None, 14, 14, 256)  0           conv4_block9_2_bn[0][0]          
__________________________________________________________________________________________________
conv4_block9_3_conv (Conv2D)    (None, 14, 14, 1024) 263168      conv4_block9_2_relu[0][0]        
__________________________________________________________________________________________________
conv4_block9_out (Add)          (None, 14, 14, 1024) 0           conv4_block8_out[0][0]           
                                                                 conv4_block9_3_conv[0][0]        
__________________________________________________________________________________________________
conv4_block10_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block9_out[0][0]           
__________________________________________________________________________________________________
conv4_block10_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block10_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block10_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block10_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block10_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block10_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block10_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block10_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block10_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block10_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block10_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block10_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block10_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block10_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block10_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block10_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block10_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block10_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block10_out (Add)         (None, 14, 14, 1024) 0           conv4_block9_out[0][0]           
                                                                 conv4_block10_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block11_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block10_out[0][0]          
__________________________________________________________________________________________________
conv4_block11_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block11_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block11_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block11_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block11_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block11_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block11_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block11_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block11_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block11_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block11_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block11_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block11_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block11_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block11_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block11_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block11_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block11_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block11_out (Add)         (None, 14, 14, 1024) 0           conv4_block10_out[0][0]          
                                                                 conv4_block11_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block12_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block11_out[0][0]          
__________________________________________________________________________________________________
conv4_block12_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block12_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block12_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block12_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block12_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block12_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block12_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block12_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block12_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block12_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block12_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block12_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block12_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block12_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block12_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block12_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block12_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block12_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block12_out (Add)         (None, 14, 14, 1024) 0           conv4_block11_out[0][0]          
                                                                 conv4_block12_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block13_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block12_out[0][0]          
__________________________________________________________________________________________________
conv4_block13_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block13_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block13_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block13_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block13_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block13_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block13_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block13_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block13_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block13_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block13_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block13_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block13_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block13_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block13_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block13_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block13_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block13_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block13_out (Add)         (None, 14, 14, 1024) 0           conv4_block12_out[0][0]          
                                                                 conv4_block13_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block14_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block13_out[0][0]          
__________________________________________________________________________________________________
conv4_block14_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block14_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block14_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block14_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block14_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block14_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block14_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block14_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block14_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block14_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block14_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block14_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block14_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block14_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block14_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block14_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block14_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block14_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block14_out (Add)         (None, 14, 14, 1024) 0           conv4_block13_out[0][0]          
                                                                 conv4_block14_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block15_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block14_out[0][0]          
__________________________________________________________________________________________________
conv4_block15_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block15_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block15_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block15_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block15_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block15_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block15_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block15_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block15_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block15_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block15_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block15_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block15_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block15_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block15_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block15_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block15_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block15_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block15_out (Add)         (None, 14, 14, 1024) 0           conv4_block14_out[0][0]          
                                                                 conv4_block15_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block16_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block15_out[0][0]          
__________________________________________________________________________________________________
conv4_block16_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block16_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block16_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block16_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block16_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block16_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block16_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block16_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block16_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block16_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block16_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block16_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block16_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block16_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block16_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block16_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block16_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block16_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block16_out (Add)         (None, 14, 14, 1024) 0           conv4_block15_out[0][0]          
                                                                 conv4_block16_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block17_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block16_out[0][0]          
__________________________________________________________________________________________________
conv4_block17_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block17_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block17_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block17_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block17_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block17_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block17_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block17_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block17_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block17_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block17_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block17_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block17_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block17_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block17_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block17_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block17_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block17_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block17_out (Add)         (None, 14, 14, 1024) 0           conv4_block16_out[0][0]          
                                                                 conv4_block17_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block18_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block17_out[0][0]          
__________________________________________________________________________________________________
conv4_block18_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block18_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block18_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block18_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block18_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block18_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block18_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block18_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block18_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block18_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block18_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block18_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block18_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block18_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block18_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block18_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block18_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block18_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block18_out (Add)         (None, 14, 14, 1024) 0           conv4_block17_out[0][0]          
                                                                 conv4_block18_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block19_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block18_out[0][0]          
__________________________________________________________________________________________________
conv4_block19_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block19_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block19_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block19_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block19_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block19_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block19_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block19_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block19_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block19_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block19_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block19_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block19_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block19_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block19_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block19_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block19_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block19_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block19_out (Add)         (None, 14, 14, 1024) 0           conv4_block18_out[0][0]          
                                                                 conv4_block19_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block20_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block19_out[0][0]          
__________________________________________________________________________________________________
conv4_block20_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block20_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block20_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block20_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block20_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block20_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block20_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block20_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block20_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block20_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block20_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block20_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block20_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block20_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block20_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block20_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block20_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block20_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block20_out (Add)         (None, 14, 14, 1024) 0           conv4_block19_out[0][0]          
                                                                 conv4_block20_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block21_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block20_out[0][0]          
__________________________________________________________________________________________________
conv4_block21_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block21_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block21_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block21_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block21_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block21_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block21_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block21_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block21_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block21_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block21_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block21_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block21_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block21_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block21_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block21_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block21_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block21_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block21_out (Add)         (None, 14, 14, 1024) 0           conv4_block20_out[0][0]          
                                                                 conv4_block21_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block22_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block21_out[0][0]          
__________________________________________________________________________________________________
conv4_block22_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block22_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block22_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block22_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block22_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block22_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block22_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block22_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block22_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block22_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block22_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block22_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block22_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block22_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block22_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block22_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block22_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block22_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block22_out (Add)         (None, 14, 14, 1024) 0           conv4_block21_out[0][0]          
                                                                 conv4_block22_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block23_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block22_out[0][0]          
__________________________________________________________________________________________________
conv4_block23_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block23_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block23_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block23_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block23_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block23_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block23_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block23_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block23_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block23_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block23_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block23_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block23_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block23_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block23_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block23_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block23_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block23_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block23_out (Add)         (None, 14, 14, 1024) 0           conv4_block22_out[0][0]          
                                                                 conv4_block23_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block24_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block23_out[0][0]          
__________________________________________________________________________________________________
conv4_block24_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block24_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block24_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block24_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block24_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block24_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block24_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block24_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block24_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block24_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block24_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block24_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block24_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block24_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block24_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block24_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block24_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block24_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block24_out (Add)         (None, 14, 14, 1024) 0           conv4_block23_out[0][0]          
                                                                 conv4_block24_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block25_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block24_out[0][0]          
__________________________________________________________________________________________________
conv4_block25_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block25_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block25_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block25_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block25_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block25_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block25_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block25_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block25_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block25_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block25_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block25_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block25_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block25_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block25_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block25_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block25_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block25_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block25_out (Add)         (None, 14, 14, 1024) 0           conv4_block24_out[0][0]          
                                                                 conv4_block25_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block26_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block25_out[0][0]          
__________________________________________________________________________________________________
conv4_block26_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block26_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block26_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block26_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block26_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block26_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block26_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block26_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block26_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block26_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block26_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block26_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block26_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block26_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block26_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block26_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block26_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block26_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block26_out (Add)         (None, 14, 14, 1024) 0           conv4_block25_out[0][0]          
                                                                 conv4_block26_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block27_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block26_out[0][0]          
__________________________________________________________________________________________________
conv4_block27_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block27_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block27_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block27_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block27_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block27_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block27_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block27_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block27_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block27_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block27_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block27_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block27_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block27_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block27_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block27_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block27_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block27_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block27_out (Add)         (None, 14, 14, 1024) 0           conv4_block26_out[0][0]          
                                                                 conv4_block27_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block28_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block27_out[0][0]          
__________________________________________________________________________________________________
conv4_block28_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block28_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block28_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block28_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block28_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block28_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block28_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block28_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block28_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block28_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block28_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block28_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block28_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block28_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block28_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block28_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block28_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block28_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block28_out (Add)         (None, 14, 14, 1024) 0           conv4_block27_out[0][0]          
                                                                 conv4_block28_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block29_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block28_out[0][0]          
__________________________________________________________________________________________________
conv4_block29_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block29_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block29_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block29_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block29_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block29_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block29_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block29_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block29_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block29_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block29_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block29_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block29_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block29_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block29_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block29_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block29_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block29_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block29_out (Add)         (None, 14, 14, 1024) 0           conv4_block28_out[0][0]          
                                                                 conv4_block29_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block30_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block29_out[0][0]          
__________________________________________________________________________________________________
conv4_block30_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block30_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block30_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block30_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block30_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block30_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block30_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block30_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block30_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block30_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block30_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block30_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block30_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block30_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block30_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block30_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block30_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block30_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block30_out (Add)         (None, 14, 14, 1024) 0           conv4_block29_out[0][0]          
                                                                 conv4_block30_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block31_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block30_out[0][0]          
__________________________________________________________________________________________________
conv4_block31_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block31_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block31_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block31_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block31_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block31_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block31_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block31_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block31_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block31_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block31_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block31_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block31_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block31_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block31_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block31_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block31_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block31_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block31_out (Add)         (None, 14, 14, 1024) 0           conv4_block30_out[0][0]          
                                                                 conv4_block31_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block32_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block31_out[0][0]          
__________________________________________________________________________________________________
conv4_block32_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block32_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block32_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block32_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block32_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block32_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block32_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block32_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block32_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block32_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block32_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block32_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block32_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block32_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block32_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block32_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block32_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block32_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block32_out (Add)         (None, 14, 14, 1024) 0           conv4_block31_out[0][0]          
                                                                 conv4_block32_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block33_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block32_out[0][0]          
__________________________________________________________________________________________________
conv4_block33_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block33_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block33_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block33_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block33_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block33_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block33_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block33_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block33_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block33_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block33_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block33_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block33_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block33_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block33_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block33_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block33_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block33_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block33_out (Add)         (None, 14, 14, 1024) 0           conv4_block32_out[0][0]          
                                                                 conv4_block33_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block34_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block33_out[0][0]          
__________________________________________________________________________________________________
conv4_block34_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block34_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block34_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block34_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block34_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block34_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block34_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block34_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block34_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block34_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block34_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block34_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block34_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block34_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block34_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block34_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block34_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block34_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block34_out (Add)         (None, 14, 14, 1024) 0           conv4_block33_out[0][0]          
                                                                 conv4_block34_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block35_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block34_out[0][0]          
__________________________________________________________________________________________________
conv4_block35_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block35_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block35_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block35_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block35_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block35_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block35_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block35_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block35_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block35_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block35_2_conv (Conv2D)   (None, 14, 14, 256)  589824      conv4_block35_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block35_2_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block35_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block35_2_relu (Activatio (None, 14, 14, 256)  0           conv4_block35_2_bn[0][0]         
__________________________________________________________________________________________________
conv4_block35_3_conv (Conv2D)   (None, 14, 14, 1024) 263168      conv4_block35_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block35_out (Add)         (None, 14, 14, 1024) 0           conv4_block34_out[0][0]          
                                                                 conv4_block35_3_conv[0][0]       
__________________________________________________________________________________________________
conv4_block36_preact_bn (BatchN (None, 14, 14, 1024) 4096        conv4_block35_out[0][0]          
__________________________________________________________________________________________________
conv4_block36_preact_relu (Acti (None, 14, 14, 1024) 0           conv4_block36_preact_bn[0][0]    
__________________________________________________________________________________________________
conv4_block36_1_conv (Conv2D)   (None, 14, 14, 256)  262144      conv4_block36_preact_relu[0][0]  
__________________________________________________________________________________________________
conv4_block36_1_bn (BatchNormal (None, 14, 14, 256)  1024        conv4_block36_1_conv[0][0]       
__________________________________________________________________________________________________
conv4_block36_1_relu (Activatio (None, 14, 14, 256)  0           conv4_block36_1_bn[0][0]         
__________________________________________________________________________________________________
conv4_block36_2_pad (ZeroPaddin (None, 16, 16, 256)  0           conv4_block36_1_relu[0][0]       
__________________________________________________________________________________________________
conv4_block36_2_conv (Conv2D)   (None, 7, 7, 256)    589824      conv4_block36_2_pad[0][0]        
__________________________________________________________________________________________________
conv4_block36_2_bn (BatchNormal (None, 7, 7, 256)    1024        conv4_block36_2_conv[0][0]       
__________________________________________________________________________________________________
conv4_block36_2_relu (Activatio (None, 7, 7, 256)    0           conv4_block36_2_bn[0][0]         
__________________________________________________________________________________________________
max_pooling2d_5 (MaxPooling2D)  (None, 7, 7, 1024)   0           conv4_block35_out[0][0]          
__________________________________________________________________________________________________
conv4_block36_3_conv (Conv2D)   (None, 7, 7, 1024)   263168      conv4_block36_2_relu[0][0]       
__________________________________________________________________________________________________
conv4_block36_out (Add)         (None, 7, 7, 1024)   0           max_pooling2d_5[0][0]            
                                                                 conv4_block36_3_conv[0][0]       
__________________________________________________________________________________________________
conv5_block1_preact_bn (BatchNo (None, 7, 7, 1024)   4096        conv4_block36_out[0][0]          
__________________________________________________________________________________________________
conv5_block1_preact_relu (Activ (None, 7, 7, 1024)   0           conv5_block1_preact_bn[0][0]     
__________________________________________________________________________________________________
conv5_block1_1_conv (Conv2D)    (None, 7, 7, 512)    524288      conv5_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv5_block1_1_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block1_1_conv[0][0]        
__________________________________________________________________________________________________
conv5_block1_1_relu (Activation (None, 7, 7, 512)    0           conv5_block1_1_bn[0][0]          
__________________________________________________________________________________________________
conv5_block1_2_pad (ZeroPadding (None, 9, 9, 512)    0           conv5_block1_1_relu[0][0]        
__________________________________________________________________________________________________
conv5_block1_2_conv (Conv2D)    (None, 7, 7, 512)    2359296     conv5_block1_2_pad[0][0]         
__________________________________________________________________________________________________
conv5_block1_2_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block1_2_conv[0][0]        
__________________________________________________________________________________________________
conv5_block1_2_relu (Activation (None, 7, 7, 512)    0           conv5_block1_2_bn[0][0]          
__________________________________________________________________________________________________
conv5_block1_0_conv (Conv2D)    (None, 7, 7, 2048)   2099200     conv5_block1_preact_relu[0][0]   
__________________________________________________________________________________________________
conv5_block1_3_conv (Conv2D)    (None, 7, 7, 2048)   1050624     conv5_block1_2_relu[0][0]        
__________________________________________________________________________________________________
conv5_block1_out (Add)          (None, 7, 7, 2048)   0           conv5_block1_0_conv[0][0]        
                                                                 conv5_block1_3_conv[0][0]        
__________________________________________________________________________________________________
conv5_block2_preact_bn (BatchNo (None, 7, 7, 2048)   8192        conv5_block1_out[0][0]           
__________________________________________________________________________________________________
conv5_block2_preact_relu (Activ (None, 7, 7, 2048)   0           conv5_block2_preact_bn[0][0]     
__________________________________________________________________________________________________
conv5_block2_1_conv (Conv2D)    (None, 7, 7, 512)    1048576     conv5_block2_preact_relu[0][0]   
__________________________________________________________________________________________________
conv5_block2_1_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block2_1_conv[0][0]        
__________________________________________________________________________________________________
conv5_block2_1_relu (Activation (None, 7, 7, 512)    0           conv5_block2_1_bn[0][0]          
__________________________________________________________________________________________________
conv5_block2_2_pad (ZeroPadding (None, 9, 9, 512)    0           conv5_block2_1_relu[0][0]        
__________________________________________________________________________________________________
conv5_block2_2_conv (Conv2D)    (None, 7, 7, 512)    2359296     conv5_block2_2_pad[0][0]         
__________________________________________________________________________________________________
conv5_block2_2_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block2_2_conv[0][0]        
__________________________________________________________________________________________________
conv5_block2_2_relu (Activation (None, 7, 7, 512)    0           conv5_block2_2_bn[0][0]          
__________________________________________________________________________________________________
conv5_block2_3_conv (Conv2D)    (None, 7, 7, 2048)   1050624     conv5_block2_2_relu[0][0]        
__________________________________________________________________________________________________
conv5_block2_out (Add)          (None, 7, 7, 2048)   0           conv5_block1_out[0][0]           
                                                                 conv5_block2_3_conv[0][0]        
__________________________________________________________________________________________________
conv5_block3_preact_bn (BatchNo (None, 7, 7, 2048)   8192        conv5_block2_out[0][0]           
__________________________________________________________________________________________________
conv5_block3_preact_relu (Activ (None, 7, 7, 2048)   0           conv5_block3_preact_bn[0][0]     
__________________________________________________________________________________________________
conv5_block3_1_conv (Conv2D)    (None, 7, 7, 512)    1048576     conv5_block3_preact_relu[0][0]   
__________________________________________________________________________________________________
conv5_block3_1_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block3_1_conv[0][0]        
__________________________________________________________________________________________________
conv5_block3_1_relu (Activation (None, 7, 7, 512)    0           conv5_block3_1_bn[0][0]          
__________________________________________________________________________________________________
conv5_block3_2_pad (ZeroPadding (None, 9, 9, 512)    0           conv5_block3_1_relu[0][0]        
__________________________________________________________________________________________________
conv5_block3_2_conv (Conv2D)    (None, 7, 7, 512)    2359296     conv5_block3_2_pad[0][0]         
__________________________________________________________________________________________________
conv5_block3_2_bn (BatchNormali (None, 7, 7, 512)    2048        conv5_block3_2_conv[0][0]        
__________________________________________________________________________________________________
conv5_block3_2_relu (Activation (None, 7, 7, 512)    0           conv5_block3_2_bn[0][0]          
__________________________________________________________________________________________________
conv5_block3_3_conv (Conv2D)    (None, 7, 7, 2048)   1050624     conv5_block3_2_relu[0][0]        
__________________________________________________________________________________________________
conv5_block3_out (Add)          (None, 7, 7, 2048)   0           conv5_block2_out[0][0]           
                                                                 conv5_block3_3_conv[0][0]        
__________________________________________________________________________________________________
post_bn (BatchNormalization)    (None, 7, 7, 2048)   8192        conv5_block3_out[0][0]           
__________________________________________________________________________________________________
post_relu (Activation)          (None, 7, 7, 2048)   0           post_bn[0][0]                    
__________________________________________________________________________________________________
avg_pool (GlobalAveragePooling2 (None, 2048)         0           post_relu[0][0]                  
__________________________________________________________________________________________________
predictions (Dense)             (None, 1000)         2049000     avg_pool[0][0]                   
==================================================================================================
Total params: 60,380,648
Trainable params: 60,236,904
Non-trainable params: 143,744
__________________________________________________________________________________________________

inception = tf.keras.applications.InceptionV3() 
inception.summary() # concatenate

Model: "inception_v3"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_3 (InputLayer)            [(None, 299, 299, 3) 0                                            
__________________________________________________________________________________________________
conv2d (Conv2D)                 (None, 149, 149, 32) 864         input_3[0][0]                    
__________________________________________________________________________________________________
batch_normalization (BatchNorma (None, 149, 149, 32) 96          conv2d[0][0]                     
__________________________________________________________________________________________________
activation (Activation)         (None, 149, 149, 32) 0           batch_normalization[0][0]        
__________________________________________________________________________________________________
conv2d_1 (Conv2D)               (None, 147, 147, 32) 9216        activation[0][0]                 
__________________________________________________________________________________________________
batch_normalization_1 (BatchNor (None, 147, 147, 32) 96          conv2d_1[0][0]                   
__________________________________________________________________________________________________
activation_1 (Activation)       (None, 147, 147, 32) 0           batch_normalization_1[0][0]      
__________________________________________________________________________________________________
conv2d_2 (Conv2D)               (None, 147, 147, 64) 18432       activation_1[0][0]               
__________________________________________________________________________________________________
batch_normalization_2 (BatchNor (None, 147, 147, 64) 192         conv2d_2[0][0]                   
__________________________________________________________________________________________________
activation_2 (Activation)       (None, 147, 147, 64) 0           batch_normalization_2[0][0]      
__________________________________________________________________________________________________
max_pooling2d_6 (MaxPooling2D)  (None, 73, 73, 64)   0           activation_2[0][0]               
__________________________________________________________________________________________________
conv2d_3 (Conv2D)               (None, 73, 73, 80)   5120        max_pooling2d_6[0][0]            
__________________________________________________________________________________________________
batch_normalization_3 (BatchNor (None, 73, 73, 80)   240         conv2d_3[0][0]                   
__________________________________________________________________________________________________
activation_3 (Activation)       (None, 73, 73, 80)   0           batch_normalization_3[0][0]      
__________________________________________________________________________________________________
conv2d_4 (Conv2D)               (None, 71, 71, 192)  138240      activation_3[0][0]               
__________________________________________________________________________________________________
batch_normalization_4 (BatchNor (None, 71, 71, 192)  576         conv2d_4[0][0]                   
__________________________________________________________________________________________________
activation_4 (Activation)       (None, 71, 71, 192)  0           batch_normalization_4[0][0]      
__________________________________________________________________________________________________
max_pooling2d_7 (MaxPooling2D)  (None, 35, 35, 192)  0           activation_4[0][0]               
__________________________________________________________________________________________________
conv2d_8 (Conv2D)               (None, 35, 35, 64)   12288       max_pooling2d_7[0][0]            
__________________________________________________________________________________________________
batch_normalization_8 (BatchNor (None, 35, 35, 64)   192         conv2d_8[0][0]                   
__________________________________________________________________________________________________
activation_8 (Activation)       (None, 35, 35, 64)   0           batch_normalization_8[0][0]      
__________________________________________________________________________________________________
conv2d_6 (Conv2D)               (None, 35, 35, 48)   9216        max_pooling2d_7[0][0]            
__________________________________________________________________________________________________
conv2d_9 (Conv2D)               (None, 35, 35, 96)   55296       activation_8[0][0]               
__________________________________________________________________________________________________
batch_normalization_6 (BatchNor (None, 35, 35, 48)   144         conv2d_6[0][0]                   
__________________________________________________________________________________________________
batch_normalization_9 (BatchNor (None, 35, 35, 96)   288         conv2d_9[0][0]                   
__________________________________________________________________________________________________
activation_6 (Activation)       (None, 35, 35, 48)   0           batch_normalization_6[0][0]      
__________________________________________________________________________________________________
activation_9 (Activation)       (None, 35, 35, 96)   0           batch_normalization_9[0][0]      
__________________________________________________________________________________________________
average_pooling2d (AveragePooli (None, 35, 35, 192)  0           max_pooling2d_7[0][0]            
__________________________________________________________________________________________________
conv2d_5 (Conv2D)               (None, 35, 35, 64)   12288       max_pooling2d_7[0][0]            
__________________________________________________________________________________________________
conv2d_7 (Conv2D)               (None, 35, 35, 64)   76800       activation_6[0][0]               
__________________________________________________________________________________________________
conv2d_10 (Conv2D)              (None, 35, 35, 96)   82944       activation_9[0][0]               
__________________________________________________________________________________________________
conv2d_11 (Conv2D)              (None, 35, 35, 32)   6144        average_pooling2d[0][0]          
__________________________________________________________________________________________________
batch_normalization_5 (BatchNor (None, 35, 35, 64)   192         conv2d_5[0][0]                   
__________________________________________________________________________________________________
batch_normalization_7 (BatchNor (None, 35, 35, 64)   192         conv2d_7[0][0]                   
__________________________________________________________________________________________________
batch_normalization_10 (BatchNo (None, 35, 35, 96)   288         conv2d_10[0][0]                  
__________________________________________________________________________________________________
batch_normalization_11 (BatchNo (None, 35, 35, 32)   96          conv2d_11[0][0]                  
__________________________________________________________________________________________________
activation_5 (Activation)       (None, 35, 35, 64)   0           batch_normalization_5[0][0]      
__________________________________________________________________________________________________
activation_7 (Activation)       (None, 35, 35, 64)   0           batch_normalization_7[0][0]      
__________________________________________________________________________________________________
activation_10 (Activation)      (None, 35, 35, 96)   0           batch_normalization_10[0][0]     
__________________________________________________________________________________________________
activation_11 (Activation)      (None, 35, 35, 32)   0           batch_normalization_11[0][0]     
__________________________________________________________________________________________________
mixed0 (Concatenate)            (None, 35, 35, 256)  0           activation_5[0][0]               
                                                                 activation_7[0][0]               
                                                                 activation_10[0][0]              
                                                                 activation_11[0][0]              
__________________________________________________________________________________________________
conv2d_15 (Conv2D)              (None, 35, 35, 64)   16384       mixed0[0][0]                     
__________________________________________________________________________________________________
batch_normalization_15 (BatchNo (None, 35, 35, 64)   192         conv2d_15[0][0]                  
__________________________________________________________________________________________________
activation_15 (Activation)      (None, 35, 35, 64)   0           batch_normalization_15[0][0]     
__________________________________________________________________________________________________
conv2d_13 (Conv2D)              (None, 35, 35, 48)   12288       mixed0[0][0]                     
__________________________________________________________________________________________________
conv2d_16 (Conv2D)              (None, 35, 35, 96)   55296       activation_15[0][0]              
__________________________________________________________________________________________________
batch_normalization_13 (BatchNo (None, 35, 35, 48)   144         conv2d_13[0][0]                  
__________________________________________________________________________________________________
batch_normalization_16 (BatchNo (None, 35, 35, 96)   288         conv2d_16[0][0]                  
__________________________________________________________________________________________________
activation_13 (Activation)      (None, 35, 35, 48)   0           batch_normalization_13[0][0]     
__________________________________________________________________________________________________
activation_16 (Activation)      (None, 35, 35, 96)   0           batch_normalization_16[0][0]     
__________________________________________________________________________________________________
average_pooling2d_1 (AveragePoo (None, 35, 35, 256)  0           mixed0[0][0]                     
__________________________________________________________________________________________________
conv2d_12 (Conv2D)              (None, 35, 35, 64)   16384       mixed0[0][0]                     
__________________________________________________________________________________________________
conv2d_14 (Conv2D)              (None, 35, 35, 64)   76800       activation_13[0][0]              
__________________________________________________________________________________________________
conv2d_17 (Conv2D)              (None, 35, 35, 96)   82944       activation_16[0][0]              
__________________________________________________________________________________________________
conv2d_18 (Conv2D)              (None, 35, 35, 64)   16384       average_pooling2d_1[0][0]        
__________________________________________________________________________________________________
batch_normalization_12 (BatchNo (None, 35, 35, 64)   192         conv2d_12[0][0]                  
__________________________________________________________________________________________________
batch_normalization_14 (BatchNo (None, 35, 35, 64)   192         conv2d_14[0][0]                  
__________________________________________________________________________________________________
batch_normalization_17 (BatchNo (None, 35, 35, 96)   288         conv2d_17[0][0]                  
__________________________________________________________________________________________________
batch_normalization_18 (BatchNo (None, 35, 35, 64)   192         conv2d_18[0][0]                  
__________________________________________________________________________________________________
activation_12 (Activation)      (None, 35, 35, 64)   0           batch_normalization_12[0][0]     
__________________________________________________________________________________________________
activation_14 (Activation)      (None, 35, 35, 64)   0           batch_normalization_14[0][0]     
__________________________________________________________________________________________________
activation_17 (Activation)      (None, 35, 35, 96)   0           batch_normalization_17[0][0]     
__________________________________________________________________________________________________
activation_18 (Activation)      (None, 35, 35, 64)   0           batch_normalization_18[0][0]     
__________________________________________________________________________________________________
mixed1 (Concatenate)            (None, 35, 35, 288)  0           activation_12[0][0]              
                                                                 activation_14[0][0]              
                                                                 activation_17[0][0]              
                                                                 activation_18[0][0]              
__________________________________________________________________________________________________
conv2d_22 (Conv2D)              (None, 35, 35, 64)   18432       mixed1[0][0]                     
__________________________________________________________________________________________________
batch_normalization_22 (BatchNo (None, 35, 35, 64)   192         conv2d_22[0][0]                  
__________________________________________________________________________________________________
activation_22 (Activation)      (None, 35, 35, 64)   0           batch_normalization_22[0][0]     
__________________________________________________________________________________________________
conv2d_20 (Conv2D)              (None, 35, 35, 48)   13824       mixed1[0][0]                     
__________________________________________________________________________________________________
conv2d_23 (Conv2D)              (None, 35, 35, 96)   55296       activation_22[0][0]              
__________________________________________________________________________________________________
batch_normalization_20 (BatchNo (None, 35, 35, 48)   144         conv2d_20[0][0]                  
__________________________________________________________________________________________________
batch_normalization_23 (BatchNo (None, 35, 35, 96)   288         conv2d_23[0][0]                  
__________________________________________________________________________________________________
activation_20 (Activation)      (None, 35, 35, 48)   0           batch_normalization_20[0][0]     
__________________________________________________________________________________________________
activation_23 (Activation)      (None, 35, 35, 96)   0           batch_normalization_23[0][0]     
__________________________________________________________________________________________________
average_pooling2d_2 (AveragePoo (None, 35, 35, 288)  0           mixed1[0][0]                     
__________________________________________________________________________________________________
conv2d_19 (Conv2D)              (None, 35, 35, 64)   18432       mixed1[0][0]                     
__________________________________________________________________________________________________
conv2d_21 (Conv2D)              (None, 35, 35, 64)   76800       activation_20[0][0]              
__________________________________________________________________________________________________
conv2d_24 (Conv2D)              (None, 35, 35, 96)   82944       activation_23[0][0]              
__________________________________________________________________________________________________
conv2d_25 (Conv2D)              (None, 35, 35, 64)   18432       average_pooling2d_2[0][0]        
__________________________________________________________________________________________________
batch_normalization_19 (BatchNo (None, 35, 35, 64)   192         conv2d_19[0][0]                  
__________________________________________________________________________________________________
batch_normalization_21 (BatchNo (None, 35, 35, 64)   192         conv2d_21[0][0]                  
__________________________________________________________________________________________________
batch_normalization_24 (BatchNo (None, 35, 35, 96)   288         conv2d_24[0][0]                  
__________________________________________________________________________________________________
batch_normalization_25 (BatchNo (None, 35, 35, 64)   192         conv2d_25[0][0]                  
__________________________________________________________________________________________________
activation_19 (Activation)      (None, 35, 35, 64)   0           batch_normalization_19[0][0]     
__________________________________________________________________________________________________
activation_21 (Activation)      (None, 35, 35, 64)   0           batch_normalization_21[0][0]     
__________________________________________________________________________________________________
activation_24 (Activation)      (None, 35, 35, 96)   0           batch_normalization_24[0][0]     
__________________________________________________________________________________________________
activation_25 (Activation)      (None, 35, 35, 64)   0           batch_normalization_25[0][0]     
__________________________________________________________________________________________________
mixed2 (Concatenate)            (None, 35, 35, 288)  0           activation_19[0][0]              
                                                                 activation_21[0][0]              
                                                                 activation_24[0][0]              
                                                                 activation_25[0][0]              
__________________________________________________________________________________________________
conv2d_27 (Conv2D)              (None, 35, 35, 64)   18432       mixed2[0][0]                     
__________________________________________________________________________________________________
batch_normalization_27 (BatchNo (None, 35, 35, 64)   192         conv2d_27[0][0]                  
__________________________________________________________________________________________________
activation_27 (Activation)      (None, 35, 35, 64)   0           batch_normalization_27[0][0]     
__________________________________________________________________________________________________
conv2d_28 (Conv2D)              (None, 35, 35, 96)   55296       activation_27[0][0]              
__________________________________________________________________________________________________
batch_normalization_28 (BatchNo (None, 35, 35, 96)   288         conv2d_28[0][0]                  
__________________________________________________________________________________________________
activation_28 (Activation)      (None, 35, 35, 96)   0           batch_normalization_28[0][0]     
__________________________________________________________________________________________________
conv2d_26 (Conv2D)              (None, 17, 17, 384)  995328      mixed2[0][0]                     
__________________________________________________________________________________________________
conv2d_29 (Conv2D)              (None, 17, 17, 96)   82944       activation_28[0][0]              
__________________________________________________________________________________________________
batch_normalization_26 (BatchNo (None, 17, 17, 384)  1152        conv2d_26[0][0]                  
__________________________________________________________________________________________________
batch_normalization_29 (BatchNo (None, 17, 17, 96)   288         conv2d_29[0][0]                  
__________________________________________________________________________________________________
activation_26 (Activation)      (None, 17, 17, 384)  0           batch_normalization_26[0][0]     
__________________________________________________________________________________________________
activation_29 (Activation)      (None, 17, 17, 96)   0           batch_normalization_29[0][0]     
__________________________________________________________________________________________________
max_pooling2d_8 (MaxPooling2D)  (None, 17, 17, 288)  0           mixed2[0][0]                     
__________________________________________________________________________________________________
mixed3 (Concatenate)            (None, 17, 17, 768)  0           activation_26[0][0]              
                                                                 activation_29[0][0]              
                                                                 max_pooling2d_8[0][0]            
__________________________________________________________________________________________________
conv2d_34 (Conv2D)              (None, 17, 17, 128)  98304       mixed3[0][0]                     
__________________________________________________________________________________________________
batch_normalization_34 (BatchNo (None, 17, 17, 128)  384         conv2d_34[0][0]                  
__________________________________________________________________________________________________
activation_34 (Activation)      (None, 17, 17, 128)  0           batch_normalization_34[0][0]     
__________________________________________________________________________________________________
conv2d_35 (Conv2D)              (None, 17, 17, 128)  114688      activation_34[0][0]              
__________________________________________________________________________________________________
batch_normalization_35 (BatchNo (None, 17, 17, 128)  384         conv2d_35[0][0]                  
__________________________________________________________________________________________________
activation_35 (Activation)      (None, 17, 17, 128)  0           batch_normalization_35[0][0]     
__________________________________________________________________________________________________
conv2d_31 (Conv2D)              (None, 17, 17, 128)  98304       mixed3[0][0]                     
__________________________________________________________________________________________________
conv2d_36 (Conv2D)              (None, 17, 17, 128)  114688      activation_35[0][0]              
__________________________________________________________________________________________________
batch_normalization_31 (BatchNo (None, 17, 17, 128)  384         conv2d_31[0][0]                  
__________________________________________________________________________________________________
batch_normalization_36 (BatchNo (None, 17, 17, 128)  384         conv2d_36[0][0]                  
__________________________________________________________________________________________________
activation_31 (Activation)      (None, 17, 17, 128)  0           batch_normalization_31[0][0]     
__________________________________________________________________________________________________
activation_36 (Activation)      (None, 17, 17, 128)  0           batch_normalization_36[0][0]     
__________________________________________________________________________________________________
conv2d_32 (Conv2D)              (None, 17, 17, 128)  114688      activation_31[0][0]              
__________________________________________________________________________________________________
conv2d_37 (Conv2D)              (None, 17, 17, 128)  114688      activation_36[0][0]              
__________________________________________________________________________________________________
batch_normalization_32 (BatchNo (None, 17, 17, 128)  384         conv2d_32[0][0]                  
__________________________________________________________________________________________________
batch_normalization_37 (BatchNo (None, 17, 17, 128)  384         conv2d_37[0][0]                  
__________________________________________________________________________________________________
activation_32 (Activation)      (None, 17, 17, 128)  0           batch_normalization_32[0][0]     
__________________________________________________________________________________________________
activation_37 (Activation)      (None, 17, 17, 128)  0           batch_normalization_37[0][0]     
__________________________________________________________________________________________________
average_pooling2d_3 (AveragePoo (None, 17, 17, 768)  0           mixed3[0][0]                     
__________________________________________________________________________________________________
conv2d_30 (Conv2D)              (None, 17, 17, 192)  147456      mixed3[0][0]                     
__________________________________________________________________________________________________
conv2d_33 (Conv2D)              (None, 17, 17, 192)  172032      activation_32[0][0]              
__________________________________________________________________________________________________
conv2d_38 (Conv2D)              (None, 17, 17, 192)  172032      activation_37[0][0]              
__________________________________________________________________________________________________
conv2d_39 (Conv2D)              (None, 17, 17, 192)  147456      average_pooling2d_3[0][0]        
__________________________________________________________________________________________________
batch_normalization_30 (BatchNo (None, 17, 17, 192)  576         conv2d_30[0][0]                  
__________________________________________________________________________________________________
batch_normalization_33 (BatchNo (None, 17, 17, 192)  576         conv2d_33[0][0]                  
__________________________________________________________________________________________________
batch_normalization_38 (BatchNo (None, 17, 17, 192)  576         conv2d_38[0][0]                  
__________________________________________________________________________________________________
batch_normalization_39 (BatchNo (None, 17, 17, 192)  576         conv2d_39[0][0]                  
__________________________________________________________________________________________________
activation_30 (Activation)      (None, 17, 17, 192)  0           batch_normalization_30[0][0]     
__________________________________________________________________________________________________
activation_33 (Activation)      (None, 17, 17, 192)  0           batch_normalization_33[0][0]     
__________________________________________________________________________________________________
activation_38 (Activation)      (None, 17, 17, 192)  0           batch_normalization_38[0][0]     
__________________________________________________________________________________________________
activation_39 (Activation)      (None, 17, 17, 192)  0           batch_normalization_39[0][0]     
__________________________________________________________________________________________________
mixed4 (Concatenate)            (None, 17, 17, 768)  0           activation_30[0][0]              
                                                                 activation_33[0][0]              
                                                                 activation_38[0][0]              
                                                                 activation_39[0][0]              
__________________________________________________________________________________________________
conv2d_44 (Conv2D)              (None, 17, 17, 160)  122880      mixed4[0][0]                     
__________________________________________________________________________________________________
batch_normalization_44 (BatchNo (None, 17, 17, 160)  480         conv2d_44[0][0]                  
__________________________________________________________________________________________________
activation_44 (Activation)      (None, 17, 17, 160)  0           batch_normalization_44[0][0]     
__________________________________________________________________________________________________
conv2d_45 (Conv2D)              (None, 17, 17, 160)  179200      activation_44[0][0]              
__________________________________________________________________________________________________
batch_normalization_45 (BatchNo (None, 17, 17, 160)  480         conv2d_45[0][0]                  
__________________________________________________________________________________________________
activation_45 (Activation)      (None, 17, 17, 160)  0           batch_normalization_45[0][0]     
__________________________________________________________________________________________________
conv2d_41 (Conv2D)              (None, 17, 17, 160)  122880      mixed4[0][0]                     
__________________________________________________________________________________________________
conv2d_46 (Conv2D)              (None, 17, 17, 160)  179200      activation_45[0][0]              
__________________________________________________________________________________________________
batch_normalization_41 (BatchNo (None, 17, 17, 160)  480         conv2d_41[0][0]                  
__________________________________________________________________________________________________
batch_normalization_46 (BatchNo (None, 17, 17, 160)  480         conv2d_46[0][0]                  
__________________________________________________________________________________________________
activation_41 (Activation)      (None, 17, 17, 160)  0           batch_normalization_41[0][0]     
__________________________________________________________________________________________________
activation_46 (Activation)      (None, 17, 17, 160)  0           batch_normalization_46[0][0]     
__________________________________________________________________________________________________
conv2d_42 (Conv2D)              (None, 17, 17, 160)  179200      activation_41[0][0]              
__________________________________________________________________________________________________
conv2d_47 (Conv2D)              (None, 17, 17, 160)  179200      activation_46[0][0]              
__________________________________________________________________________________________________
batch_normalization_42 (BatchNo (None, 17, 17, 160)  480         conv2d_42[0][0]                  
__________________________________________________________________________________________________
batch_normalization_47 (BatchNo (None, 17, 17, 160)  480         conv2d_47[0][0]                  
__________________________________________________________________________________________________
activation_42 (Activation)      (None, 17, 17, 160)  0           batch_normalization_42[0][0]     
__________________________________________________________________________________________________
activation_47 (Activation)      (None, 17, 17, 160)  0           batch_normalization_47[0][0]     
__________________________________________________________________________________________________
average_pooling2d_4 (AveragePoo (None, 17, 17, 768)  0           mixed4[0][0]                     
__________________________________________________________________________________________________
conv2d_40 (Conv2D)              (None, 17, 17, 192)  147456      mixed4[0][0]                     
__________________________________________________________________________________________________
conv2d_43 (Conv2D)              (None, 17, 17, 192)  215040      activation_42[0][0]              
__________________________________________________________________________________________________
conv2d_48 (Conv2D)              (None, 17, 17, 192)  215040      activation_47[0][0]              
__________________________________________________________________________________________________
conv2d_49 (Conv2D)              (None, 17, 17, 192)  147456      average_pooling2d_4[0][0]        
__________________________________________________________________________________________________
batch_normalization_40 (BatchNo (None, 17, 17, 192)  576         conv2d_40[0][0]                  
__________________________________________________________________________________________________
batch_normalization_43 (BatchNo (None, 17, 17, 192)  576         conv2d_43[0][0]                  
__________________________________________________________________________________________________
batch_normalization_48 (BatchNo (None, 17, 17, 192)  576         conv2d_48[0][0]                  
__________________________________________________________________________________________________
batch_normalization_49 (BatchNo (None, 17, 17, 192)  576         conv2d_49[0][0]                  
__________________________________________________________________________________________________
activation_40 (Activation)      (None, 17, 17, 192)  0           batch_normalization_40[0][0]     
__________________________________________________________________________________________________
activation_43 (Activation)      (None, 17, 17, 192)  0           batch_normalization_43[0][0]     
__________________________________________________________________________________________________
activation_48 (Activation)      (None, 17, 17, 192)  0           batch_normalization_48[0][0]     
__________________________________________________________________________________________________
activation_49 (Activation)      (None, 17, 17, 192)  0           batch_normalization_49[0][0]     
__________________________________________________________________________________________________
mixed5 (Concatenate)            (None, 17, 17, 768)  0           activation_40[0][0]              
                                                                 activation_43[0][0]              
                                                                 activation_48[0][0]              
                                                                 activation_49[0][0]              
__________________________________________________________________________________________________
conv2d_54 (Conv2D)              (None, 17, 17, 160)  122880      mixed5[0][0]                     
__________________________________________________________________________________________________
batch_normalization_54 (BatchNo (None, 17, 17, 160)  480         conv2d_54[0][0]                  
__________________________________________________________________________________________________
activation_54 (Activation)      (None, 17, 17, 160)  0           batch_normalization_54[0][0]     
__________________________________________________________________________________________________
conv2d_55 (Conv2D)              (None, 17, 17, 160)  179200      activation_54[0][0]              
__________________________________________________________________________________________________
batch_normalization_55 (BatchNo (None, 17, 17, 160)  480         conv2d_55[0][0]                  
__________________________________________________________________________________________________
activation_55 (Activation)      (None, 17, 17, 160)  0           batch_normalization_55[0][0]     
__________________________________________________________________________________________________
conv2d_51 (Conv2D)              (None, 17, 17, 160)  122880      mixed5[0][0]                     
__________________________________________________________________________________________________
conv2d_56 (Conv2D)              (None, 17, 17, 160)  179200      activation_55[0][0]              
__________________________________________________________________________________________________
batch_normalization_51 (BatchNo (None, 17, 17, 160)  480         conv2d_51[0][0]                  
__________________________________________________________________________________________________
batch_normalization_56 (BatchNo (None, 17, 17, 160)  480         conv2d_56[0][0]                  
__________________________________________________________________________________________________
activation_51 (Activation)      (None, 17, 17, 160)  0           batch_normalization_51[0][0]     
__________________________________________________________________________________________________
activation_56 (Activation)      (None, 17, 17, 160)  0           batch_normalization_56[0][0]     
__________________________________________________________________________________________________
conv2d_52 (Conv2D)              (None, 17, 17, 160)  179200      activation_51[0][0]              
__________________________________________________________________________________________________
conv2d_57 (Conv2D)              (None, 17, 17, 160)  179200      activation_56[0][0]              
__________________________________________________________________________________________________
batch_normalization_52 (BatchNo (None, 17, 17, 160)  480         conv2d_52[0][0]                  
__________________________________________________________________________________________________
batch_normalization_57 (BatchNo (None, 17, 17, 160)  480         conv2d_57[0][0]                  
__________________________________________________________________________________________________
activation_52 (Activation)      (None, 17, 17, 160)  0           batch_normalization_52[0][0]     
__________________________________________________________________________________________________
activation_57 (Activation)      (None, 17, 17, 160)  0           batch_normalization_57[0][0]     
__________________________________________________________________________________________________
average_pooling2d_5 (AveragePoo (None, 17, 17, 768)  0           mixed5[0][0]                     
__________________________________________________________________________________________________
conv2d_50 (Conv2D)              (None, 17, 17, 192)  147456      mixed5[0][0]                     
__________________________________________________________________________________________________
conv2d_53 (Conv2D)              (None, 17, 17, 192)  215040      activation_52[0][0]              
__________________________________________________________________________________________________
conv2d_58 (Conv2D)              (None, 17, 17, 192)  215040      activation_57[0][0]              
__________________________________________________________________________________________________
conv2d_59 (Conv2D)              (None, 17, 17, 192)  147456      average_pooling2d_5[0][0]        
__________________________________________________________________________________________________
batch_normalization_50 (BatchNo (None, 17, 17, 192)  576         conv2d_50[0][0]                  
__________________________________________________________________________________________________
batch_normalization_53 (BatchNo (None, 17, 17, 192)  576         conv2d_53[0][0]                  
__________________________________________________________________________________________________
batch_normalization_58 (BatchNo (None, 17, 17, 192)  576         conv2d_58[0][0]                  
__________________________________________________________________________________________________
batch_normalization_59 (BatchNo (None, 17, 17, 192)  576         conv2d_59[0][0]                  
__________________________________________________________________________________________________
activation_50 (Activation)      (None, 17, 17, 192)  0           batch_normalization_50[0][0]     
__________________________________________________________________________________________________
activation_53 (Activation)      (None, 17, 17, 192)  0           batch_normalization_53[0][0]     
__________________________________________________________________________________________________
activation_58 (Activation)      (None, 17, 17, 192)  0           batch_normalization_58[0][0]     
__________________________________________________________________________________________________
activation_59 (Activation)      (None, 17, 17, 192)  0           batch_normalization_59[0][0]     
__________________________________________________________________________________________________
mixed6 (Concatenate)            (None, 17, 17, 768)  0           activation_50[0][0]              
                                                                 activation_53[0][0]              
                                                                 activation_58[0][0]              
                                                                 activation_59[0][0]              
__________________________________________________________________________________________________
conv2d_64 (Conv2D)              (None, 17, 17, 192)  147456      mixed6[0][0]                     
__________________________________________________________________________________________________
batch_normalization_64 (BatchNo (None, 17, 17, 192)  576         conv2d_64[0][0]                  
__________________________________________________________________________________________________
activation_64 (Activation)      (None, 17, 17, 192)  0           batch_normalization_64[0][0]     
__________________________________________________________________________________________________
conv2d_65 (Conv2D)              (None, 17, 17, 192)  258048      activation_64[0][0]              
__________________________________________________________________________________________________
batch_normalization_65 (BatchNo (None, 17, 17, 192)  576         conv2d_65[0][0]                  
__________________________________________________________________________________________________
activation_65 (Activation)      (None, 17, 17, 192)  0           batch_normalization_65[0][0]     
__________________________________________________________________________________________________
conv2d_61 (Conv2D)              (None, 17, 17, 192)  147456      mixed6[0][0]                     
__________________________________________________________________________________________________
conv2d_66 (Conv2D)              (None, 17, 17, 192)  258048      activation_65[0][0]              
__________________________________________________________________________________________________
batch_normalization_61 (BatchNo (None, 17, 17, 192)  576         conv2d_61[0][0]                  
__________________________________________________________________________________________________
batch_normalization_66 (BatchNo (None, 17, 17, 192)  576         conv2d_66[0][0]                  
__________________________________________________________________________________________________
activation_61 (Activation)      (None, 17, 17, 192)  0           batch_normalization_61[0][0]     
__________________________________________________________________________________________________
activation_66 (Activation)      (None, 17, 17, 192)  0           batch_normalization_66[0][0]     
__________________________________________________________________________________________________
conv2d_62 (Conv2D)              (None, 17, 17, 192)  258048      activation_61[0][0]              
__________________________________________________________________________________________________
conv2d_67 (Conv2D)              (None, 17, 17, 192)  258048      activation_66[0][0]              
__________________________________________________________________________________________________
batch_normalization_62 (BatchNo (None, 17, 17, 192)  576         conv2d_62[0][0]                  
__________________________________________________________________________________________________
batch_normalization_67 (BatchNo (None, 17, 17, 192)  576         conv2d_67[0][0]                  
__________________________________________________________________________________________________
activation_62 (Activation)      (None, 17, 17, 192)  0           batch_normalization_62[0][0]     
__________________________________________________________________________________________________
activation_67 (Activation)      (None, 17, 17, 192)  0           batch_normalization_67[0][0]     
__________________________________________________________________________________________________
average_pooling2d_6 (AveragePoo (None, 17, 17, 768)  0           mixed6[0][0]                     
__________________________________________________________________________________________________
conv2d_60 (Conv2D)              (None, 17, 17, 192)  147456      mixed6[0][0]                     
__________________________________________________________________________________________________
conv2d_63 (Conv2D)              (None, 17, 17, 192)  258048      activation_62[0][0]              
__________________________________________________________________________________________________
conv2d_68 (Conv2D)              (None, 17, 17, 192)  258048      activation_67[0][0]              
__________________________________________________________________________________________________
conv2d_69 (Conv2D)              (None, 17, 17, 192)  147456      average_pooling2d_6[0][0]        
__________________________________________________________________________________________________
batch_normalization_60 (BatchNo (None, 17, 17, 192)  576         conv2d_60[0][0]                  
__________________________________________________________________________________________________
batch_normalization_63 (BatchNo (None, 17, 17, 192)  576         conv2d_63[0][0]                  
__________________________________________________________________________________________________
batch_normalization_68 (BatchNo (None, 17, 17, 192)  576         conv2d_68[0][0]                  
__________________________________________________________________________________________________
batch_normalization_69 (BatchNo (None, 17, 17, 192)  576         conv2d_69[0][0]                  
__________________________________________________________________________________________________
activation_60 (Activation)      (None, 17, 17, 192)  0           batch_normalization_60[0][0]     
__________________________________________________________________________________________________
activation_63 (Activation)      (None, 17, 17, 192)  0           batch_normalization_63[0][0]     
__________________________________________________________________________________________________
activation_68 (Activation)      (None, 17, 17, 192)  0           batch_normalization_68[0][0]     
__________________________________________________________________________________________________
activation_69 (Activation)      (None, 17, 17, 192)  0           batch_normalization_69[0][0]     
__________________________________________________________________________________________________
mixed7 (Concatenate)            (None, 17, 17, 768)  0           activation_60[0][0]              
                                                                 activation_63[0][0]              
                                                                 activation_68[0][0]              
                                                                 activation_69[0][0]              
__________________________________________________________________________________________________
conv2d_72 (Conv2D)              (None, 17, 17, 192)  147456      mixed7[0][0]                     
__________________________________________________________________________________________________
batch_normalization_72 (BatchNo (None, 17, 17, 192)  576         conv2d_72[0][0]                  
__________________________________________________________________________________________________
activation_72 (Activation)      (None, 17, 17, 192)  0           batch_normalization_72[0][0]     
__________________________________________________________________________________________________
conv2d_73 (Conv2D)              (None, 17, 17, 192)  258048      activation_72[0][0]              
__________________________________________________________________________________________________
batch_normalization_73 (BatchNo (None, 17, 17, 192)  576         conv2d_73[0][0]                  
__________________________________________________________________________________________________
activation_73 (Activation)      (None, 17, 17, 192)  0           batch_normalization_73[0][0]     
__________________________________________________________________________________________________
conv2d_70 (Conv2D)              (None, 17, 17, 192)  147456      mixed7[0][0]                     
__________________________________________________________________________________________________
conv2d_74 (Conv2D)              (None, 17, 17, 192)  258048      activation_73[0][0]              
__________________________________________________________________________________________________
batch_normalization_70 (BatchNo (None, 17, 17, 192)  576         conv2d_70[0][0]                  
__________________________________________________________________________________________________
batch_normalization_74 (BatchNo (None, 17, 17, 192)  576         conv2d_74[0][0]                  
__________________________________________________________________________________________________
activation_70 (Activation)      (None, 17, 17, 192)  0           batch_normalization_70[0][0]     
__________________________________________________________________________________________________
activation_74 (Activation)      (None, 17, 17, 192)  0           batch_normalization_74[0][0]     
__________________________________________________________________________________________________
conv2d_71 (Conv2D)              (None, 8, 8, 320)    552960      activation_70[0][0]              
__________________________________________________________________________________________________
conv2d_75 (Conv2D)              (None, 8, 8, 192)    331776      activation_74[0][0]              
__________________________________________________________________________________________________
batch_normalization_71 (BatchNo (None, 8, 8, 320)    960         conv2d_71[0][0]                  
__________________________________________________________________________________________________
batch_normalization_75 (BatchNo (None, 8, 8, 192)    576         conv2d_75[0][0]                  
__________________________________________________________________________________________________
activation_71 (Activation)      (None, 8, 8, 320)    0           batch_normalization_71[0][0]     
__________________________________________________________________________________________________
activation_75 (Activation)      (None, 8, 8, 192)    0           batch_normalization_75[0][0]     
__________________________________________________________________________________________________
max_pooling2d_9 (MaxPooling2D)  (None, 8, 8, 768)    0           mixed7[0][0]                     
__________________________________________________________________________________________________
mixed8 (Concatenate)            (None, 8, 8, 1280)   0           activation_71[0][0]              
                                                                 activation_75[0][0]              
                                                                 max_pooling2d_9[0][0]            
__________________________________________________________________________________________________
conv2d_80 (Conv2D)              (None, 8, 8, 448)    573440      mixed8[0][0]                     
__________________________________________________________________________________________________
batch_normalization_80 (BatchNo (None, 8, 8, 448)    1344        conv2d_80[0][0]                  
__________________________________________________________________________________________________
activation_80 (Activation)      (None, 8, 8, 448)    0           batch_normalization_80[0][0]     
__________________________________________________________________________________________________
conv2d_77 (Conv2D)              (None, 8, 8, 384)    491520      mixed8[0][0]                     
__________________________________________________________________________________________________
conv2d_81 (Conv2D)              (None, 8, 8, 384)    1548288     activation_80[0][0]              
__________________________________________________________________________________________________
batch_normalization_77 (BatchNo (None, 8, 8, 384)    1152        conv2d_77[0][0]                  
__________________________________________________________________________________________________
batch_normalization_81 (BatchNo (None, 8, 8, 384)    1152        conv2d_81[0][0]                  
__________________________________________________________________________________________________
activation_77 (Activation)      (None, 8, 8, 384)    0           batch_normalization_77[0][0]     
__________________________________________________________________________________________________
activation_81 (Activation)      (None, 8, 8, 384)    0           batch_normalization_81[0][0]     
__________________________________________________________________________________________________
conv2d_78 (Conv2D)              (None, 8, 8, 384)    442368      activation_77[0][0]              
__________________________________________________________________________________________________
conv2d_79 (Conv2D)              (None, 8, 8, 384)    442368      activation_77[0][0]              
__________________________________________________________________________________________________
conv2d_82 (Conv2D)              (None, 8, 8, 384)    442368      activation_81[0][0]              
__________________________________________________________________________________________________
conv2d_83 (Conv2D)              (None, 8, 8, 384)    442368      activation_81[0][0]              
__________________________________________________________________________________________________
average_pooling2d_7 (AveragePoo (None, 8, 8, 1280)   0           mixed8[0][0]                     
__________________________________________________________________________________________________
conv2d_76 (Conv2D)              (None, 8, 8, 320)    409600      mixed8[0][0]                     
__________________________________________________________________________________________________
batch_normalization_78 (BatchNo (None, 8, 8, 384)    1152        conv2d_78[0][0]                  
__________________________________________________________________________________________________
batch_normalization_79 (BatchNo (None, 8, 8, 384)    1152        conv2d_79[0][0]                  
__________________________________________________________________________________________________
batch_normalization_82 (BatchNo (None, 8, 8, 384)    1152        conv2d_82[0][0]                  
__________________________________________________________________________________________________
batch_normalization_83 (BatchNo (None, 8, 8, 384)    1152        conv2d_83[0][0]                  
__________________________________________________________________________________________________
conv2d_84 (Conv2D)              (None, 8, 8, 192)    245760      average_pooling2d_7[0][0]        
__________________________________________________________________________________________________
batch_normalization_76 (BatchNo (None, 8, 8, 320)    960         conv2d_76[0][0]                  
__________________________________________________________________________________________________
activation_78 (Activation)      (None, 8, 8, 384)    0           batch_normalization_78[0][0]     
__________________________________________________________________________________________________
activation_79 (Activation)      (None, 8, 8, 384)    0           batch_normalization_79[0][0]     
__________________________________________________________________________________________________
activation_82 (Activation)      (None, 8, 8, 384)    0           batch_normalization_82[0][0]     
__________________________________________________________________________________________________
activation_83 (Activation)      (None, 8, 8, 384)    0           batch_normalization_83[0][0]     
__________________________________________________________________________________________________
batch_normalization_84 (BatchNo (None, 8, 8, 192)    576         conv2d_84[0][0]                  
__________________________________________________________________________________________________
activation_76 (Activation)      (None, 8, 8, 320)    0           batch_normalization_76[0][0]     
__________________________________________________________________________________________________
mixed9_0 (Concatenate)          (None, 8, 8, 768)    0           activation_78[0][0]              
                                                                 activation_79[0][0]              
__________________________________________________________________________________________________
concatenate (Concatenate)       (None, 8, 8, 768)    0           activation_82[0][0]              
                                                                 activation_83[0][0]              
__________________________________________________________________________________________________
activation_84 (Activation)      (None, 8, 8, 192)    0           batch_normalization_84[0][0]     
__________________________________________________________________________________________________
mixed9 (Concatenate)            (None, 8, 8, 2048)   0           activation_76[0][0]              
                                                                 mixed9_0[0][0]                   
                                                                 concatenate[0][0]                
                                                                 activation_84[0][0]              
__________________________________________________________________________________________________
conv2d_89 (Conv2D)              (None, 8, 8, 448)    917504      mixed9[0][0]                     
__________________________________________________________________________________________________
batch_normalization_89 (BatchNo (None, 8, 8, 448)    1344        conv2d_89[0][0]                  
__________________________________________________________________________________________________
activation_89 (Activation)      (None, 8, 8, 448)    0           batch_normalization_89[0][0]     
__________________________________________________________________________________________________
conv2d_86 (Conv2D)              (None, 8, 8, 384)    786432      mixed9[0][0]                     
__________________________________________________________________________________________________
conv2d_90 (Conv2D)              (None, 8, 8, 384)    1548288     activation_89[0][0]              
__________________________________________________________________________________________________
batch_normalization_86 (BatchNo (None, 8, 8, 384)    1152        conv2d_86[0][0]                  
__________________________________________________________________________________________________
batch_normalization_90 (BatchNo (None, 8, 8, 384)    1152        conv2d_90[0][0]                  
__________________________________________________________________________________________________
activation_86 (Activation)      (None, 8, 8, 384)    0           batch_normalization_86[0][0]     
__________________________________________________________________________________________________
activation_90 (Activation)      (None, 8, 8, 384)    0           batch_normalization_90[0][0]     
__________________________________________________________________________________________________
conv2d_87 (Conv2D)              (None, 8, 8, 384)    442368      activation_86[0][0]              
__________________________________________________________________________________________________
conv2d_88 (Conv2D)              (None, 8, 8, 384)    442368      activation_86[0][0]              
__________________________________________________________________________________________________
conv2d_91 (Conv2D)              (None, 8, 8, 384)    442368      activation_90[0][0]              
__________________________________________________________________________________________________
conv2d_92 (Conv2D)              (None, 8, 8, 384)    442368      activation_90[0][0]              
__________________________________________________________________________________________________
average_pooling2d_8 (AveragePoo (None, 8, 8, 2048)   0           mixed9[0][0]                     
__________________________________________________________________________________________________
conv2d_85 (Conv2D)              (None, 8, 8, 320)    655360      mixed9[0][0]                     
__________________________________________________________________________________________________
batch_normalization_87 (BatchNo (None, 8, 8, 384)    1152        conv2d_87[0][0]                  
__________________________________________________________________________________________________
batch_normalization_88 (BatchNo (None, 8, 8, 384)    1152        conv2d_88[0][0]                  
__________________________________________________________________________________________________
batch_normalization_91 (BatchNo (None, 8, 8, 384)    1152        conv2d_91[0][0]                  
__________________________________________________________________________________________________
batch_normalization_92 (BatchNo (None, 8, 8, 384)    1152        conv2d_92[0][0]                  
__________________________________________________________________________________________________
conv2d_93 (Conv2D)              (None, 8, 8, 192)    393216      average_pooling2d_8[0][0]        
__________________________________________________________________________________________________
batch_normalization_85 (BatchNo (None, 8, 8, 320)    960         conv2d_85[0][0]                  
__________________________________________________________________________________________________
activation_87 (Activation)      (None, 8, 8, 384)    0           batch_normalization_87[0][0]     
__________________________________________________________________________________________________
activation_88 (Activation)      (None, 8, 8, 384)    0           batch_normalization_88[0][0]     
__________________________________________________________________________________________________
activation_91 (Activation)      (None, 8, 8, 384)    0           batch_normalization_91[0][0]     
__________________________________________________________________________________________________
activation_92 (Activation)      (None, 8, 8, 384)    0           batch_normalization_92[0][0]     
__________________________________________________________________________________________________
batch_normalization_93 (BatchNo (None, 8, 8, 192)    576         conv2d_93[0][0]                  
__________________________________________________________________________________________________
activation_85 (Activation)      (None, 8, 8, 320)    0           batch_normalization_85[0][0]     
__________________________________________________________________________________________________
mixed9_1 (Concatenate)          (None, 8, 8, 768)    0           activation_87[0][0]              
                                                                 activation_88[0][0]              
__________________________________________________________________________________________________
concatenate_1 (Concatenate)     (None, 8, 8, 768)    0           activation_91[0][0]              
                                                                 activation_92[0][0]              
__________________________________________________________________________________________________
activation_93 (Activation)      (None, 8, 8, 192)    0           batch_normalization_93[0][0]     
__________________________________________________________________________________________________
mixed10 (Concatenate)           (None, 8, 8, 2048)   0           activation_85[0][0]              
                                                                 mixed9_1[0][0]                   
                                                                 concatenate_1[0][0]              
                                                                 activation_93[0][0]              
__________________________________________________________________________________________________
avg_pool (GlobalAveragePooling2 (None, 2048)         0           mixed10[0][0]                    
__________________________________________________________________________________________________
predictions (Dense)             (None, 1000)         2049000     avg_pool[0][0]                   
==================================================================================================
Total params: 23,851,784
Trainable params: 23,817,352
Non-trainable params: 34,432
__________________________________________________________________________________________________

concatenate는 원본 값을 그대로 유지하고 싶을 때 (구조를 합침)
add는 합쳐서 하나의 결과를 유지한다

원본 이미지와 조건을 더할때 일반적으로 concatenate를 사용하는 것이 유용한다
add를 사용해야 할때는 concatenate를 하고나서도 해도 되기 때문이다

import numpy as np

a = np.array([[1,2,3]])
b = np.array([[1,2,3]])

a+b
array([[2, 4, 6]])

c = np.concatenate((a,b))

c
# array([[1, 2, 3],
#        [1, 2, 3]])

c.sum(axis=0) 
# array([2, 4, 6])

cGAN의 목표

이러한 과정을 통해 G는 입력 이미지 x와 noise z로 부터 출력 이미지 y로 매핑하도록 학습이 된다

Pix2Pix 구현

import tensorflow as tf

import os
import pathlib
import time
import datetime

from matplotlib import pyplot as plt
from IPython import display

dataset_name = "facades"

_URL = f'http://efrosgans.eecs.berkeley.edu/pix2pix/datasets/{dataset_name}.tar.gz'

path_to_zip = tf.keras.utils.get_file(
    fname=f"{dataset_name}.tar.gz",
    origin=_URL,
    extract=True)

path_to_zip  = pathlib.Path(path_to_zip)

PATH = path_to_zip.parent/dataset_name

list(PATH.parent.iterdir())

[PosixPath('/Users/jihyeokjeong/.keras/datasets/imdb_word_index.json'),
 PosixPath('/Users/jihyeokjeong/.keras/datasets/mnist.npz'),
 PosixPath('/Users/jihyeokjeong/.keras/datasets/fashion-mnist'),
 PosixPath('/Users/jihyeokjeong/.keras/datasets/imdb.npz'),
 PosixPath('/Users/jihyeokjeong/.keras/datasets/facades'),
 PosixPath('/Users/jihyeokjeong/.keras/datasets/facades.tar.gz')]

sample_image = tf.io.read_file(str(PATH / 'train/1.jpg'))
sample_image = tf.io.decode_jpeg(sample_image)
print(sample_image.shape)
# (256, 512, 3)

plt.figure()
plt.imshow(sample_image)

def load(image_file):
  image = tf.io.read_file(image_file)
  image = tf.image.decode_jpeg(image)
    
  w = tf.shape(image)[1]
  w = w // 2
  input_image = image[:, w:, :]
  real_image = image[:, :w, :]

  input_image = tf.cast(input_image, tf.float32)
  real_image = tf.cast(real_image, tf.float32)

  return input_image, real_image

inp, re = load(str(PATH / 'train/100.jpg'))

plt.figure()
plt.imshow(inp / 255.0)
plt.figure()
plt.imshow(re / 255.0)

BUFFER_SIZE = 400
BATCH_SIZE = 1
IMG_WIDTH = 256
IMG_HEIGHT = 256

def resize(input_image, real_image, height, width):
  input_image = tf.image.resize(input_image, [height, width],
                                method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)
  real_image = tf.image.resize(real_image, [height, width],
                               method=tf.image.ResizeMethod.NEAREST_NEIGHBOR)

  return input_image, real_image

def random_crop(input_image, real_image):
  stacked_image = tf.stack([input_image, real_image], axis=0)
  cropped_image = tf.image.random_crop(
      stacked_image, size=[2, IMG_HEIGHT, IMG_WIDTH, 3])

  return cropped_image[0], cropped_image[1]

def normalize(input_image, real_image):
  input_image = (input_image / 127.5) - 1
  real_image = (real_image / 127.5) - 1

  return input_image, real_image

@tf.function()
def random_jitter(input_image, real_image):
  input_image, real_image = resize(input_image, real_image, 286, 286)

  input_image, real_image = random_crop(input_image, real_image)

  if tf.random.uniform(()) > 0.5:
    input_image = tf.image.flip_left_right(input_image)
    real_image = tf.image.flip_left_right(real_image)

  return input_image, real_image

plt.figure(figsize=(6, 6))
for i in range(4):
  rj_inp, rj_re = random_jitter(inp, re)
  plt.subplot(2, 2, i + 1)
  plt.imshow(rj_inp / 255.0)
  plt.axis('off')
plt.show()

def load_image_train(image_file):
  input_image, real_image = load(image_file)
  input_image, real_image = random_jitter(input_image, real_image)
  input_image, real_image = normalize(input_image, real_image)

  return input_image, real_image

def load_image_test(image_file):
  input_image, real_image = load(image_file)
  input_image, real_image = resize(input_image, real_image,
                                   IMG_HEIGHT, IMG_WIDTH)
  input_image, real_image = normalize(input_image, real_image)

  return input_image, real_image

train_dataset = tf.data.Dataset.list_files(str(PATH / 'train/*.jpg'))
train_dataset = train_dataset.map(load_image_train,
                                  num_parallel_calls=tf.data.AUTOTUNE)
train_dataset = train_dataset.shuffle(BUFFER_SIZE)
train_dataset = train_dataset.batch(BATCH_SIZE)

try:
  test_dataset = tf.data.Dataset.list_files(str(PATH / 'test/*.jpg'))
except tf.errors.InvalidArgumentError:
  test_dataset = tf.data.Dataset.list_files(str(PATH / 'val/*.jpg'))
test_dataset = test_dataset.map(load_image_test)
test_dataset = test_dataset.batch(BATCH_SIZE)

OUTPUT_CHANNELS = 3

def downsample(filters, size, apply_batchnorm=True):
  initializer = tf.random_normal_initializer(0., 0.02)

  result = tf.keras.Sequential()
  result.add(
      tf.keras.layers.Conv2D(filters, size, strides=2, padding='same',
                             kernel_initializer=initializer, use_bias=False)) # bias 사용 옵션

  if apply_batchnorm:
    result.add(tf.keras.layers.BatchNormalization()) # BM는 옵션

  result.add(tf.keras.layers.LeakyReLU())

  return result

down_model = downsample(3, 4)
down_result = down_model(tf.expand_dims(inp, 0))
print (down_result.shape)
# (1, 128, 128, 3)

def upsample(filters, size, apply_dropout=False):
  initializer = tf.random_normal_initializer(0., 0.02)

  result = tf.keras.Sequential()
  result.add(
    tf.keras.layers.Conv2DTranspose(filters, size, strides=2,
                                    padding='same',
                                    kernel_initializer=initializer,
                                    use_bias=False)) 
  # bias를 사용하지 않는 이유 : GAN은 기본적으로 학습이 잘 안되기 때문에 간소화 시켜야 하므로 bias를 사용하지 않는다
  
  result.add(tf.keras.layers.BatchNormalization())
  
  # layer가 많을 때는 dropout을 사용하지 않아도 되기 때문에 옵션으로 사용 
  # 범용적 사용을 위해 옵션으로 만듦
  if apply_dropout:
      result.add(tf.keras.layers.Dropout(0.5))

  result.add(tf.keras.layers.ReLU())
  
  return result

up_model = upsample(3, 4)
up_result = up_model(down_result)
print (up_result.shape)
# (1, 256, 256, 3)

# Generator안에서 U-net 생성
# U-net을 가짜 데이터 생성용으로 사용
def Generator():
  inputs = tf.keras.layers.Input(shape=[256, 256, 3])

  down_stack = [
    # U-net구조를 위해 size확인을 해야 한다
    downsample(64, 4, apply_batchnorm=False),  # (batch_size, 128, 128, 64)
    downsample(128, 4),  # (batch_size, 64, 64, 128)
    downsample(256, 4),  # (batch_size, 32, 32, 256)
    downsample(512, 4),  # (batch_size, 16, 16, 512)
    downsample(512, 4),  # (batch_size, 8, 8, 512)
    downsample(512, 4),  # (batch_size, 4, 4, 512)
    downsample(512, 4),  # (batch_size, 2, 2, 512)
    downsample(512, 4),  # (batch_size, 1, 1, 512)
  ]

  up_stack = [
    upsample(512, 4, apply_dropout=True),  # (batch_size, 2, 2, 1024)
    upsample(512, 4, apply_dropout=True),  # (batch_size, 4, 4, 1024)
    upsample(512, 4, apply_dropout=True),  # (batch_size, 8, 8, 1024)
    upsample(512, 4),  # (batch_size, 16, 16, 1024)
    upsample(256, 4),  # (batch_size, 32, 32, 512)
    upsample(128, 4),  # (batch_size, 64, 64, 256)
    upsample(64, 4),  # (batch_size, 128, 128, 128)
  ]

  initializer = tf.random_normal_initializer(0., 0.02)
  last = tf.keras.layers.Conv2DTranspose(OUTPUT_CHANNELS, 4,
                                         strides=2,
                                         padding='same',
                                         kernel_initializer=initializer,
                                         activation='tanh')  # (batch_size, 256, 256, 3)
                            # sigmoid를 사용하면 zero centered가 되지 않아 학습이 잘 안되기 때문에 tanh를 사용한다
                            # zero centered가 되지 않으면 약간의 변화에도 값이 크게 변하기 때문에 학습이 잘 안될 수 있다 
  x = inputs

  skips = []
  for down in down_stack:
    x = down(x)
    skips.append(x)

  skips = reversed(skips[:-1])

  for up, skip in zip(up_stack, skips):
    x = up(x)
    x = tf.keras.layers.Concatenate()([x, skip])

  x = last(x)

  return tf.keras.Model(inputs=inputs, outputs=x)

AutoEncoder, U-net 구조는 loss를 어떻게 하느냐에 따라, 학습 데이터를 어떻게 하느냐에 따라 목적이 달라질 수 있다
U-net구조를 통해 segmentation을 사용하기도 하고, 원본 이미지를 복원하기도 하고, 원본 이미지의 색상을 바꾸기도 한다

generator = Generator()

gen_output = generator(inp[tf.newaxis, ...], training=False)
plt.imshow(gen_output[0, ...])

LAMBDA = 100
# tanh을 사용했기 때문에 BinaryCrossentropy을 사용
loss_object = tf.keras.losses.BinaryCrossentropy(from_logits=True)

# One-sided label smoothing: 실제 데이터에 대한 target 값을 1보다 약간 작은 값으로 설정하는 방식 
# 여기서는 one-sided label smoothing 방식을 사용하지 않았다
def generator_loss(disc_generated_output, gen_output, target):
  gan_loss = loss_object(tf.ones_like(disc_generated_output), disc_generated_output)

  l1_loss = tf.reduce_mean(tf.abs(target - gen_output))

  total_gen_loss = gan_loss + (LAMBDA * l1_loss)

  return total_gen_loss, gan_loss, l1_loss

def Discriminator():
  initializer = tf.random_normal_initializer(0., 0.02)

  inp = tf.keras.layers.Input(shape=[256, 256, 3], name='input_image')
  tar = tf.keras.layers.Input(shape=[256, 256, 3], name='target_image')
  
  # 실제값 + 목표 이미지 concatenate
  x = tf.keras.layers.concatenate([inp, tar])  # (batch_size, 256, 256, channels*2)

  down1 = downsample(64, 4, False)(x)  # (batch_size, 128, 128, 64)
  down2 = downsample(128, 4)(down1)  # (batch_size, 64, 64, 128)
  down3 = downsample(256, 4)(down2)  # (batch_size, 32, 32, 256)

  # 출력의 각 30x30 크기 이미지 패치는 입력 이미지의 70x70부분을 판별한다 (부분으로 쪼개어 판별한다/사이즈는 실험을 통해 알아내야 한다)
  # 크기 맞추는 방법을 사용한다 
  zero_pad1 = tf.keras.layers.ZeroPadding2D()(down3)  # (batch_size, 34, 34, 256)
  conv = tf.keras.layers.Conv2D(512, 4, strides=1,
                                kernel_initializer=initializer,
                                use_bias=False)(zero_pad1)  # (batch_size, 31, 31, 512)

  batchnorm1 = tf.keras.layers.BatchNormalization()(conv)

  leaky_relu = tf.keras.layers.LeakyReLU()(batchnorm1)

  zero_pad2 = tf.keras.layers.ZeroPadding2D()(leaky_relu)  # (batch_size, 33, 33, 512)

  last = tf.keras.layers.Conv2D(1, 4, strides=1,
                                kernel_initializer=initializer)(zero_pad2)  # (batch_size, 30, 30, 1)

  return tf.keras.Model(inputs=[inp, tar], outputs=last)

discriminator = Discriminator()

disc_out = discriminator([inp[tf.newaxis, ...], gen_output], training=False)
plt.imshow(disc_out[0, ..., -1], vmin=-20, vmax=20, cmap='RdBu_r')
plt.colorbar()

def discriminator_loss(disc_real_output, disc_generated_output):
  # 진짜를 진짜로 1
  real_loss = loss_object(tf.ones_like(disc_real_output), disc_real_output)
  
  # 가짜는 가짜로 0
  generated_loss = loss_object(tf.zeros_like(disc_generated_output), disc_generated_output)

  total_disc_loss = real_loss + generated_loss

  return total_disc_loss

generator_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)
discriminator_optimizer = tf.keras.optimizers.Adam(2e-4, beta_1=0.5)

checkpoint_dir = './training_checkpoints'
checkpoint_prefix = os.path.join(checkpoint_dir, "ckpt")
checkpoint = tf.train.Checkpoint(generator_optimizer=generator_optimizer,
                                 discriminator_optimizer=discriminator_optimizer,
                                 generator=generator,
                                 discriminator=discriminator)

def generate_images(model, test_input, tar):
  prediction = model(test_input, training=True)
  plt.figure(figsize=(15, 15))

  display_list = [test_input[0], tar[0], prediction[0]]
  title = ['Input Image', 'Ground Truth', 'Predicted Image']

  for i in range(3):
    plt.subplot(1, 3, i+1)
    plt.title(title[i])
    plt.imshow(display_list[i] * 0.5 + 0.5)
    plt.axis('off')
  plt.show()

for example_input, example_target in test_dataset.take(1):
  generate_images(generator, example_input, example_target)

log_dir="logs/"

summary_writer = tf.summary.create_file_writer(
  log_dir + "fit/" + datetime.datetime.now().strftime("%Y%m%d-%H%M%S"))

@tf.function
def train_step(input_image, target, step):
  # generator가 먼저 실행되지만 discriminator가 먼저 전체가 실행된다 (with 중첩 때문에)
  # discriminator 학습 -> generator 학습 -> discriminator 학습 -> generator 학습 반복 
  with tf.GradientTape() as gen_tape, tf.GradientTape() as disc_tape:
    # 미분할 대상을 지정한다
    gen_output = generator(input_image, training=True)

    disc_real_output = discriminator([input_image, target], training=True)
    disc_generated_output = discriminator([input_image, gen_output], training=True)

    gen_total_loss, gen_gan_loss, gen_l1_loss = generator_loss(disc_generated_output, gen_output, target)
    disc_loss = discriminator_loss(disc_real_output, disc_generated_output)

  generator_gradients = gen_tape.gradient(gen_total_loss,
                                          generator.trainable_variables)
  discriminator_gradients = disc_tape.gradient(disc_loss,
                                               discriminator.trainable_variables)

  generator_optimizer.apply_gradients(zip(generator_gradients,
                                          generator.trainable_variables))
  discriminator_optimizer.apply_gradients(zip(discriminator_gradients,
                                              discriminator.trainable_variables))

  with summary_writer.as_default():
    tf.summary.scalar('gen_total_loss', gen_total_loss, step=step//1000)
    tf.summary.scalar('gen_gan_loss', gen_gan_loss, step=step//1000)
    tf.summary.scalar('gen_l1_loss', gen_l1_loss, step=step//1000)
    tf.summary.scalar('disc_loss', disc_loss, step=step//1000)

with A:
    with B:
        X()
    Y()

1. A __enter__
2. B __enter__
3. X()
4. B __exit__
5. Y()
6. A __exit__

def fit(train_ds, test_ds, steps):
  example_input, example_target = next(iter(test_ds.take(1)))
  start = time.time()

  for step, (input_image, target) in train_ds.repeat().take(steps).enumerate():
    if (step) % 1000 == 0:
      display.clear_output(wait=True)

      if step != 0:
        print(f'Time taken for 1000 steps: {time.time()-start:.2f} sec\n')

      start = time.time()

      generate_images(generator, example_input, example_target)
      print(f"Step: {step//1000}k")

    train_step(input_image, target, step)

    if (step+1) % 10 == 0:
      print('.', end='', flush=True)

    if (step + 1) % 5000 == 0:
      checkpoint.save(file_prefix=checkpoint_prefix)

# 총 weight 업데이트를 4만번 (GAN은 기본적으로 학습이 잘 안되기 때문에 Epochs을 높게 잡아야 한다)
fit(train_dataset, test_dataset, steps=40000)
# Time taken for 1000 steps: 159.76 sec

for inp, tar in test_dataset.take(5):
  generate_images(generator, inp, tar)

for inp, tar in test_dataset.take(6):
  generate_images(generator, inp, tar)

저작자표시 비영리

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

38일차 - Style Transfer (0)	2021.11.16
37일차 - 논문 수업 (Segmentation) (0)	2021.11.16
36일차 - 논문 수업 (Segmentation) (0)	2021.11.16
35일차 - 논문 수업 (Segmentation) (0)	2021.11.01
34일차 - 논문 수업 (VAE) (0)	2021.11.01

38일차 - Style Transfer

2021. 11. 16. 19:21

728x90

Transfer learning 정리

1. Feature extraction
- CNN (Convolution 연산 부분만)을 활용하여 추출한다 => window 연산을 하기 때문에 입력 개수가 정해지지 않아도 된다
- input_shape을 정하지 않아도 되지만 tensorflow/keras에서 build하기 위해 고정하는 경우가 있다
- Feature extracion을 사용하는 경우 전체를 쓰는 것보다 성능이 일반적으로 좋지 않다. 하지만 데이터가 적을 때 최소한 학습한 내용이 있기 때문에 조금의 영향력을 가질 수 있다 (fine tuning이 필요하다)
2. Fine tuning
- Learning rate를 크게하면 catastropic forgetting이 발생되기 때문에 learning rate를 작게 하여 재학습 시켜야 한다

U-net 구현하기

import tensorflow as tf 
import matplotlib.pyplot as plt
import numpy as np

base_model = tf.keras.applications.MobileNetV2(include_top=False)

base_model.summary()
Model: "mobilenetv2_1.00_224"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_12 (InputLayer)           [(None, None, None,  0                                            
__________________________________________________________________________________________________
Conv1 (Conv2D)                  (None, None, None, 3 864         input_12[0][0]                   
__________________________________________________________________________________________________
bn_Conv1 (BatchNormalization)   (None, None, None, 3 128         Conv1[0][0]                      
__________________________________________________________________________________________________
Conv1_relu (ReLU)               (None, None, None, 3 0           bn_Conv1[0][0]                   
__________________________________________________________________________________________________
expanded_conv_depthwise (Depthw (None, None, None, 3 288         Conv1_relu[0][0]                 
__________________________________________________________________________________________________
expanded_conv_depthwise_BN (Bat (None, None, None, 3 128         expanded_conv_depthwise[0][0]    
__________________________________________________________________________________________________
expanded_conv_depthwise_relu (R (None, None, None, 3 0           expanded_conv_depthwise_BN[0][0] 
__________________________________________________________________________________________________
expanded_conv_project (Conv2D)  (None, None, None, 1 512         expanded_conv_depthwise_relu[0][0
__________________________________________________________________________________________________
expanded_conv_project_BN (Batch (None, None, None, 1 64          expanded_conv_project[0][0]      
__________________________________________________________________________________________________
block_1_expand (Conv2D)         (None, None, None, 9 1536        expanded_conv_project_BN[0][0]   
__________________________________________________________________________________________________
block_1_expand_BN (BatchNormali (None, None, None, 9 384         block_1_expand[0][0]             
__________________________________________________________________________________________________
block_1_expand_relu (ReLU)      (None, None, None, 9 0           block_1_expand_BN[0][0]          
__________________________________________________________________________________________________
block_1_pad (ZeroPadding2D)     (None, None, None, 9 0           block_1_expand_relu[0][0]        
__________________________________________________________________________________________________
block_1_depthwise (DepthwiseCon (None, None, None, 9 864         block_1_pad[0][0]                
__________________________________________________________________________________________________
block_1_depthwise_BN (BatchNorm (None, None, None, 9 384         block_1_depthwise[0][0]          
__________________________________________________________________________________________________
block_1_depthwise_relu (ReLU)   (None, None, None, 9 0           block_1_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_1_project (Conv2D)        (None, None, None, 2 2304        block_1_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_1_project_BN (BatchNormal (None, None, None, 2 96          block_1_project[0][0]            
__________________________________________________________________________________________________
block_2_expand (Conv2D)         (None, None, None, 1 3456        block_1_project_BN[0][0]         
__________________________________________________________________________________________________
block_2_expand_BN (BatchNormali (None, None, None, 1 576         block_2_expand[0][0]             
__________________________________________________________________________________________________
block_2_expand_relu (ReLU)      (None, None, None, 1 0           block_2_expand_BN[0][0]          
__________________________________________________________________________________________________
block_2_depthwise (DepthwiseCon (None, None, None, 1 1296        block_2_expand_relu[0][0]        
__________________________________________________________________________________________________
block_2_depthwise_BN (BatchNorm (None, None, None, 1 576         block_2_depthwise[0][0]          
__________________________________________________________________________________________________
block_2_depthwise_relu (ReLU)   (None, None, None, 1 0           block_2_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_2_project (Conv2D)        (None, None, None, 2 3456        block_2_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_2_project_BN (BatchNormal (None, None, None, 2 96          block_2_project[0][0]            
__________________________________________________________________________________________________
block_2_add (Add)               (None, None, None, 2 0           block_1_project_BN[0][0]         
                                                                 block_2_project_BN[0][0]         
__________________________________________________________________________________________________
block_3_expand (Conv2D)         (None, None, None, 1 3456        block_2_add[0][0]                
__________________________________________________________________________________________________
block_3_expand_BN (BatchNormali (None, None, None, 1 576         block_3_expand[0][0]             
__________________________________________________________________________________________________
block_3_expand_relu (ReLU)      (None, None, None, 1 0           block_3_expand_BN[0][0]          
__________________________________________________________________________________________________
block_3_pad (ZeroPadding2D)     (None, None, None, 1 0           block_3_expand_relu[0][0]        
__________________________________________________________________________________________________
block_3_depthwise (DepthwiseCon (None, None, None, 1 1296        block_3_pad[0][0]                
__________________________________________________________________________________________________
block_3_depthwise_BN (BatchNorm (None, None, None, 1 576         block_3_depthwise[0][0]          
__________________________________________________________________________________________________
block_3_depthwise_relu (ReLU)   (None, None, None, 1 0           block_3_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_3_project (Conv2D)        (None, None, None, 3 4608        block_3_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_3_project_BN (BatchNormal (None, None, None, 3 128         block_3_project[0][0]            
__________________________________________________________________________________________________
block_4_expand (Conv2D)         (None, None, None, 1 6144        block_3_project_BN[0][0]         
__________________________________________________________________________________________________
block_4_expand_BN (BatchNormali (None, None, None, 1 768         block_4_expand[0][0]             
__________________________________________________________________________________________________
block_4_expand_relu (ReLU)      (None, None, None, 1 0           block_4_expand_BN[0][0]          
__________________________________________________________________________________________________
block_4_depthwise (DepthwiseCon (None, None, None, 1 1728        block_4_expand_relu[0][0]        
__________________________________________________________________________________________________
block_4_depthwise_BN (BatchNorm (None, None, None, 1 768         block_4_depthwise[0][0]          
__________________________________________________________________________________________________
block_4_depthwise_relu (ReLU)   (None, None, None, 1 0           block_4_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_4_project (Conv2D)        (None, None, None, 3 6144        block_4_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_4_project_BN (BatchNormal (None, None, None, 3 128         block_4_project[0][0]            
__________________________________________________________________________________________________
block_4_add (Add)               (None, None, None, 3 0           block_3_project_BN[0][0]         
                                                                 block_4_project_BN[0][0]         
__________________________________________________________________________________________________
block_5_expand (Conv2D)         (None, None, None, 1 6144        block_4_add[0][0]                
__________________________________________________________________________________________________
block_5_expand_BN (BatchNormali (None, None, None, 1 768         block_5_expand[0][0]             
__________________________________________________________________________________________________
block_5_expand_relu (ReLU)      (None, None, None, 1 0           block_5_expand_BN[0][0]          
__________________________________________________________________________________________________
block_5_depthwise (DepthwiseCon (None, None, None, 1 1728        block_5_expand_relu[0][0]        
__________________________________________________________________________________________________
block_5_depthwise_BN (BatchNorm (None, None, None, 1 768         block_5_depthwise[0][0]          
__________________________________________________________________________________________________
block_5_depthwise_relu (ReLU)   (None, None, None, 1 0           block_5_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_5_project (Conv2D)        (None, None, None, 3 6144        block_5_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_5_project_BN (BatchNormal (None, None, None, 3 128         block_5_project[0][0]            
__________________________________________________________________________________________________
block_5_add (Add)               (None, None, None, 3 0           block_4_add[0][0]                
                                                                 block_5_project_BN[0][0]         
__________________________________________________________________________________________________
block_6_expand (Conv2D)         (None, None, None, 1 6144        block_5_add[0][0]                
__________________________________________________________________________________________________
block_6_expand_BN (BatchNormali (None, None, None, 1 768         block_6_expand[0][0]             
__________________________________________________________________________________________________
block_6_expand_relu (ReLU)      (None, None, None, 1 0           block_6_expand_BN[0][0]          
__________________________________________________________________________________________________
block_6_pad (ZeroPadding2D)     (None, None, None, 1 0           block_6_expand_relu[0][0]        
__________________________________________________________________________________________________
block_6_depthwise (DepthwiseCon (None, None, None, 1 1728        block_6_pad[0][0]                
__________________________________________________________________________________________________
block_6_depthwise_BN (BatchNorm (None, None, None, 1 768         block_6_depthwise[0][0]          
__________________________________________________________________________________________________
block_6_depthwise_relu (ReLU)   (None, None, None, 1 0           block_6_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_6_project (Conv2D)        (None, None, None, 6 12288       block_6_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_6_project_BN (BatchNormal (None, None, None, 6 256         block_6_project[0][0]            
__________________________________________________________________________________________________
block_7_expand (Conv2D)         (None, None, None, 3 24576       block_6_project_BN[0][0]         
__________________________________________________________________________________________________
block_7_expand_BN (BatchNormali (None, None, None, 3 1536        block_7_expand[0][0]             
__________________________________________________________________________________________________
block_7_expand_relu (ReLU)      (None, None, None, 3 0           block_7_expand_BN[0][0]          
__________________________________________________________________________________________________
block_7_depthwise (DepthwiseCon (None, None, None, 3 3456        block_7_expand_relu[0][0]        
__________________________________________________________________________________________________
block_7_depthwise_BN (BatchNorm (None, None, None, 3 1536        block_7_depthwise[0][0]          
__________________________________________________________________________________________________
block_7_depthwise_relu (ReLU)   (None, None, None, 3 0           block_7_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_7_project (Conv2D)        (None, None, None, 6 24576       block_7_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_7_project_BN (BatchNormal (None, None, None, 6 256         block_7_project[0][0]            
__________________________________________________________________________________________________
block_7_add (Add)               (None, None, None, 6 0           block_6_project_BN[0][0]         
                                                                 block_7_project_BN[0][0]         
__________________________________________________________________________________________________
block_8_expand (Conv2D)         (None, None, None, 3 24576       block_7_add[0][0]                
__________________________________________________________________________________________________
block_8_expand_BN (BatchNormali (None, None, None, 3 1536        block_8_expand[0][0]             
__________________________________________________________________________________________________
block_8_expand_relu (ReLU)      (None, None, None, 3 0           block_8_expand_BN[0][0]          
__________________________________________________________________________________________________
block_8_depthwise (DepthwiseCon (None, None, None, 3 3456        block_8_expand_relu[0][0]        
__________________________________________________________________________________________________
block_8_depthwise_BN (BatchNorm (None, None, None, 3 1536        block_8_depthwise[0][0]          
__________________________________________________________________________________________________
block_8_depthwise_relu (ReLU)   (None, None, None, 3 0           block_8_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_8_project (Conv2D)        (None, None, None, 6 24576       block_8_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_8_project_BN (BatchNormal (None, None, None, 6 256         block_8_project[0][0]            
__________________________________________________________________________________________________
block_8_add (Add)               (None, None, None, 6 0           block_7_add[0][0]                
                                                                 block_8_project_BN[0][0]         
__________________________________________________________________________________________________
block_9_expand (Conv2D)         (None, None, None, 3 24576       block_8_add[0][0]                
__________________________________________________________________________________________________
block_9_expand_BN (BatchNormali (None, None, None, 3 1536        block_9_expand[0][0]             
__________________________________________________________________________________________________
block_9_expand_relu (ReLU)      (None, None, None, 3 0           block_9_expand_BN[0][0]          
__________________________________________________________________________________________________
block_9_depthwise (DepthwiseCon (None, None, None, 3 3456        block_9_expand_relu[0][0]        
__________________________________________________________________________________________________
block_9_depthwise_BN (BatchNorm (None, None, None, 3 1536        block_9_depthwise[0][0]          
__________________________________________________________________________________________________
block_9_depthwise_relu (ReLU)   (None, None, None, 3 0           block_9_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_9_project (Conv2D)        (None, None, None, 6 24576       block_9_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_9_project_BN (BatchNormal (None, None, None, 6 256         block_9_project[0][0]            
__________________________________________________________________________________________________
block_9_add (Add)               (None, None, None, 6 0           block_8_add[0][0]                
                                                                 block_9_project_BN[0][0]         
__________________________________________________________________________________________________
block_10_expand (Conv2D)        (None, None, None, 3 24576       block_9_add[0][0]                
__________________________________________________________________________________________________
block_10_expand_BN (BatchNormal (None, None, None, 3 1536        block_10_expand[0][0]            
__________________________________________________________________________________________________
block_10_expand_relu (ReLU)     (None, None, None, 3 0           block_10_expand_BN[0][0]         
__________________________________________________________________________________________________
block_10_depthwise (DepthwiseCo (None, None, None, 3 3456        block_10_expand_relu[0][0]       
__________________________________________________________________________________________________
block_10_depthwise_BN (BatchNor (None, None, None, 3 1536        block_10_depthwise[0][0]         
__________________________________________________________________________________________________
block_10_depthwise_relu (ReLU)  (None, None, None, 3 0           block_10_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_10_project (Conv2D)       (None, None, None, 9 36864       block_10_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_10_project_BN (BatchNorma (None, None, None, 9 384         block_10_project[0][0]           
__________________________________________________________________________________________________
block_11_expand (Conv2D)        (None, None, None, 5 55296       block_10_project_BN[0][0]        
__________________________________________________________________________________________________
block_11_expand_BN (BatchNormal (None, None, None, 5 2304        block_11_expand[0][0]            
__________________________________________________________________________________________________
block_11_expand_relu (ReLU)     (None, None, None, 5 0           block_11_expand_BN[0][0]         
__________________________________________________________________________________________________
block_11_depthwise (DepthwiseCo (None, None, None, 5 5184        block_11_expand_relu[0][0]       
__________________________________________________________________________________________________
block_11_depthwise_BN (BatchNor (None, None, None, 5 2304        block_11_depthwise[0][0]         
__________________________________________________________________________________________________
block_11_depthwise_relu (ReLU)  (None, None, None, 5 0           block_11_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_11_project (Conv2D)       (None, None, None, 9 55296       block_11_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_11_project_BN (BatchNorma (None, None, None, 9 384         block_11_project[0][0]           
__________________________________________________________________________________________________
block_11_add (Add)              (None, None, None, 9 0           block_10_project_BN[0][0]        
                                                                 block_11_project_BN[0][0]        
__________________________________________________________________________________________________
block_12_expand (Conv2D)        (None, None, None, 5 55296       block_11_add[0][0]               
__________________________________________________________________________________________________
block_12_expand_BN (BatchNormal (None, None, None, 5 2304        block_12_expand[0][0]            
__________________________________________________________________________________________________
block_12_expand_relu (ReLU)     (None, None, None, 5 0           block_12_expand_BN[0][0]         
__________________________________________________________________________________________________
block_12_depthwise (DepthwiseCo (None, None, None, 5 5184        block_12_expand_relu[0][0]       
__________________________________________________________________________________________________
block_12_depthwise_BN (BatchNor (None, None, None, 5 2304        block_12_depthwise[0][0]         
__________________________________________________________________________________________________
block_12_depthwise_relu (ReLU)  (None, None, None, 5 0           block_12_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_12_project (Conv2D)       (None, None, None, 9 55296       block_12_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_12_project_BN (BatchNorma (None, None, None, 9 384         block_12_project[0][0]           
__________________________________________________________________________________________________
block_12_add (Add)              (None, None, None, 9 0           block_11_add[0][0]               
                                                                 block_12_project_BN[0][0]        
__________________________________________________________________________________________________
block_13_expand (Conv2D)        (None, None, None, 5 55296       block_12_add[0][0]               
__________________________________________________________________________________________________
block_13_expand_BN (BatchNormal (None, None, None, 5 2304        block_13_expand[0][0]            
__________________________________________________________________________________________________
block_13_expand_relu (ReLU)     (None, None, None, 5 0           block_13_expand_BN[0][0]         
__________________________________________________________________________________________________
block_13_pad (ZeroPadding2D)    (None, None, None, 5 0           block_13_expand_relu[0][0]       
__________________________________________________________________________________________________
block_13_depthwise (DepthwiseCo (None, None, None, 5 5184        block_13_pad[0][0]               
__________________________________________________________________________________________________
block_13_depthwise_BN (BatchNor (None, None, None, 5 2304        block_13_depthwise[0][0]         
__________________________________________________________________________________________________
block_13_depthwise_relu (ReLU)  (None, None, None, 5 0           block_13_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_13_project (Conv2D)       (None, None, None, 1 92160       block_13_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_13_project_BN (BatchNorma (None, None, None, 1 640         block_13_project[0][0]           
__________________________________________________________________________________________________
block_14_expand (Conv2D)        (None, None, None, 9 153600      block_13_project_BN[0][0]        
__________________________________________________________________________________________________
block_14_expand_BN (BatchNormal (None, None, None, 9 3840        block_14_expand[0][0]            
__________________________________________________________________________________________________
block_14_expand_relu (ReLU)     (None, None, None, 9 0           block_14_expand_BN[0][0]         
__________________________________________________________________________________________________
block_14_depthwise (DepthwiseCo (None, None, None, 9 8640        block_14_expand_relu[0][0]       
__________________________________________________________________________________________________
block_14_depthwise_BN (BatchNor (None, None, None, 9 3840        block_14_depthwise[0][0]         
__________________________________________________________________________________________________
block_14_depthwise_relu (ReLU)  (None, None, None, 9 0           block_14_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_14_project (Conv2D)       (None, None, None, 1 153600      block_14_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_14_project_BN (BatchNorma (None, None, None, 1 640         block_14_project[0][0]           
__________________________________________________________________________________________________
block_14_add (Add)              (None, None, None, 1 0           block_13_project_BN[0][0]        
                                                                 block_14_project_BN[0][0]        
__________________________________________________________________________________________________
block_15_expand (Conv2D)        (None, None, None, 9 153600      block_14_add[0][0]               
__________________________________________________________________________________________________
block_15_expand_BN (BatchNormal (None, None, None, 9 3840        block_15_expand[0][0]            
__________________________________________________________________________________________________
block_15_expand_relu (ReLU)     (None, None, None, 9 0           block_15_expand_BN[0][0]         
__________________________________________________________________________________________________
block_15_depthwise (DepthwiseCo (None, None, None, 9 8640        block_15_expand_relu[0][0]       
__________________________________________________________________________________________________
block_15_depthwise_BN (BatchNor (None, None, None, 9 3840        block_15_depthwise[0][0]         
__________________________________________________________________________________________________
block_15_depthwise_relu (ReLU)  (None, None, None, 9 0           block_15_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_15_project (Conv2D)       (None, None, None, 1 153600      block_15_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_15_project_BN (BatchNorma (None, None, None, 1 640         block_15_project[0][0]           
__________________________________________________________________________________________________
block_15_add (Add)              (None, None, None, 1 0           block_14_add[0][0]               
                                                                 block_15_project_BN[0][0]        
__________________________________________________________________________________________________
block_16_expand (Conv2D)        (None, None, None, 9 153600      block_15_add[0][0]               
__________________________________________________________________________________________________
block_16_expand_BN (BatchNormal (None, None, None, 9 3840        block_16_expand[0][0]            
__________________________________________________________________________________________________
block_16_expand_relu (ReLU)     (None, None, None, 9 0           block_16_expand_BN[0][0]         
__________________________________________________________________________________________________
block_16_depthwise (DepthwiseCo (None, None, None, 9 8640        block_16_expand_relu[0][0]       
__________________________________________________________________________________________________
block_16_depthwise_BN (BatchNor (None, None, None, 9 3840        block_16_depthwise[0][0]         
__________________________________________________________________________________________________
block_16_depthwise_relu (ReLU)  (None, None, None, 9 0           block_16_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_16_project (Conv2D)       (None, None, None, 3 307200      block_16_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_16_project_BN (BatchNorma (None, None, None, 3 1280        block_16_project[0][0]           
__________________________________________________________________________________________________
Conv_1 (Conv2D)                 (None, None, None, 1 409600      block_16_project_BN[0][0]        
__________________________________________________________________________________________________
Conv_1_bn (BatchNormalization)  (None, None, None, 1 5120        Conv_1[0][0]                     
__________________________________________________________________________________________________
out_relu (ReLU)                 (None, None, None, 1 0           Conv_1_bn[0][0]                  
==================================================================================================
Total params: 2,257,984
Trainable params: 2,223,872
Non-trainable params: 34,112
__________________________________________________________________________________________________

im = tf.keras.preprocessing.image.load_img('ade.jpg')

np.array(im).shape
# (960, 1280, 3)

plt.imshow(im)

# base_model은 input shape이 224,224,3을 받도록 만들어져 있지만 
# 크기를 고정하지 않았기 때문에 input data로 들어갈 수 있다
base_model(np.array(im)[np.newaxis])

<tf.Tensor: shape=(1, 30, 40, 1280), dtype=float32, numpy=
array([[[[0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ]],

        [[0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.02680304],
         [0.        , 0.        , 0.        , ..., 0.        ,
          2.0374901 , 1.3772459 ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          1.1740499 , 0.43281242],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ]],

        [[0.        , 0.7047172 , 0.        , ..., 0.        ,
          1.3220952 , 2.8031015 ],
         [0.        , 0.1590641 , 0.        , ..., 0.        ,
          2.0702174 , 3.4394553 ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          1.5714682 , 2.23192   ],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.3797317 ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ]],

        ...,

        [[0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.6268955 ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.09846598]],

        [[0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.70790523, 0.28814578, 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.20785622, 0.        , ..., 0.        ,
          0.        , 0.        ],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ]],

        [[0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         ...,
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ],
         [0.        , 0.        , 0.        , ..., 0.        ,
          0.        , 0.        ]]]], dtype=float32)>

# input_shape 고정 / mobilenet(pre-trained model)은 데이터가 적을 때 그나마 괜찮은 성능을 보여준다
base_model = tf.keras.applications.MobileNetV2(input_shape=(128,128,3), include_top=False)

base_model(np.array(im)[np.newaxis]) # input_shape크기를 고정했기 때문에 입력값으로 들어가지 않는다

layer_names = [
    'block_1_expand_relu', # 64x64
    'block_3_expand_relu', # 32x32
    'block_6_expand_relu', # 16x16
    'block_13_expand_relu', # 8x8
    'block_16_project', # 4x4
]
base_model_outputs = [base_model.get_layer(name).output for name in layer_names]

base_model_outputs

[<KerasTensor: shape=(None, 64, 64, 96) dtype=float32 (created by layer 'block_1_expand_relu')>,
 <KerasTensor: shape=(None, 32, 32, 144) dtype=float32 (created by layer 'block_3_expand_relu')>,
 <KerasTensor: shape=(None, 16, 16, 192) dtype=float32 (created by layer 'block_6_expand_relu')>,
 <KerasTensor: shape=(None, 8, 8, 576) dtype=float32 (created by layer 'block_13_expand_relu')>,
 <KerasTensor: shape=(None, 4, 4, 320) dtype=float32 (created by layer 'block_16_project')>]

down_stack = tf.keras.Model(inputs=base_model.input, outputs=base_model_outputs)
down_stack.trainable = False

Concatenate vs Add

- Concatenate는 구조를 더하는 것 (구조를 더하면 정보가 그대로 유지된다)
- Add는 값을 더하는 것

GAN(Generative Adversarial Network)

GAN 구조

Generator(생성자)와 Discriminator (판별자) 두 개의 모델이 동시에 적대적인 과정으로 학습한다
생성자 G는 실제 데이터 분포를 학습하고, 판별자 D는 원래의 데이터인지 생성자로부터 생성이 된 것인지 구분한다
생성자 G의 학습 과정은 이미지를 잘 생성해서 속일 확률을 높이고 판별자 D가 제대로 구분하는 확률을 높이는 방향으로 학습을 진행한다

GAN은 진짜 같은 가짜 데이터를 generator가 생성하고 진짜 데이터와 비교하여 discriminator가 어느 것이 진짜인지 판별하며 학습해 나가는 모델이다

GAN의 종류

(X_train, y_train), (X_test, y_test) = tf.keras.datasets.mnist.load_data()

X_train = X_train.reshape(-1,28*28) / 255
X_test = X_test.reshape(-1,28*28) / 255

input_ = tf.keras.Input((10,))
x = tf.keras.layers.Dense(128)(input_)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(256)(x)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(512)(x)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(28*28, activation='sigmoid')(x)

generator = tf.keras.models.Model(input_, x)

input_ = tf.keras.Input((784,))
x = tf.keras.layers.Dense(1024)(input_)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(512)(x)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(256)(x)
x = tf.keras.layers.LeakyReLU()(x)
x = tf.keras.layers.Dense(1, activation='sigmoid')(x)

discriminator = tf.keras.models.Model(input_, x)
discriminator.compile(loss='binary_crossentropy',optimizer='adam')

discriminator.trainable = False

gan_input = tf.keras.Input((10,))
x = generator(gan_input)
output = discriminator(x)

GAN = tf.keras.models.Model(gan_input, output)
GAN.compile(loss='binary_crossentropy',optimizer='adam')

GAN.summary()

Model: "model_2"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_3 (InputLayer)         [(None, 10)]              0         
_________________________________________________________________
model (Functional)           (None, 784)               568208    
_________________________________________________________________
model_1 (Functional)         (None, 1)                 1460225   
=================================================================
Total params: 2,028,433
Trainable params: 568,208
Non-trainable params: 1,460,225
_________________________________________________________________

def get_batch(data, batch_size=32):
    batches = []
    for i in range(int(data.shape[0]//batch_size)):
        batch = data[i*batch_size: (i+1)*batch_size]
        batches.append(batch)
    return np.asarray(batches)
    

get_batch(X_train, 100).shape
#(600, 100, 784)

d_losses = []
g_losses = []
for i in range(1,11):
    for j in get_batch(X_train):
        input_noise = np.random.uniform(-1,1, size=[32,10])
        fake = generator.predict(input_noise)
        x_dis = np.concatenate([j, fake])
        y_dis = np.zeros(2*32)
        y_dis[:32] = 1
        
        discriminator.trainable = True 
        d_loss = discriminator.train_on_batch(x_dis, y_dis) # fit 
        discriminator.trainable = False 
        
        noise = np.random.uniform(-1,1,size=[32,10])
        y_gan = np.ones(32)
        g_loss = GAN.train_on_batch(noise, y_gan)
        
    d_losses.append(d_loss)
    g_losses.append(g_loss)

fake = generator.predict(noise).reshape(-1, 28,28)

plt.imshow(fake[0]) # 한 번 학습 했을 때

plt.imshow(fake[1]) # 한 번 학습 했을 때

plt.imshow(fake[0]) # 두 번 학습 했을 때

plt.imshow(fake[1]) # 두 번 학습 했을 때

Train_on_batch

Train_on_batch는 고정된 batch size를 신경쓰지 않고 현재 전달받은 입력 데이터를 활용하여 weight를 업데이트한다
GAN에서는 매번 Generator가 새로운 fake image를 만들기 때문에 epoch마다 새로운 데이터를 넘겨줘야 한다
따라서 train_on_batch를 사용하는 것이 좋다
또한 pre-trained 모델을 단일 배치로 학습시켜야 할 때 train_on_batch를 사용하는 것이 좋다

저작자표시 비영리

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

39일차 - 논문 수업 (GAN) (0)	2021.11.16
37일차 - 논문 수업 (Segmentation) (0)	2021.11.16
36일차 - 논문 수업 (Segmentation) (0)	2021.11.16
35일차 - 논문 수업 (Segmentation) (0)	2021.11.01
34일차 - 논문 수업 (VAE) (0)	2021.11.01

37일차 - 논문 수업 (Segmentation)

2021. 11. 16. 18:56

728x90

Segmentation

이미지에서 점 단위로 예측하는 기법

확인해야 할 포인트
1. 어떻게 한점 한점을 예측할지
2. 원본의 크기와 같아야 한다
- 입력된 이미지를 넣었을 때 원본과 같은 크기로 나와야 한다

종류

1. Semantic segmentation
- 이미지 내에 있는 물체들을 의미 있는 단위로 분할한다
- 이미지의 각 픽셀이 어느 클래스에 속하는지 예측한다
- 동일한 객체들끼리는 한 번에 Masking을 수행한다
- 대표적으로 FCN이 있다
2. Instance segmentation
- 동일한 객체들이여도 개별 Object로 간주하고 Masking을 수행한다
- 대표적으로 Mask R-CNN이 있다

FCN(Fully Convolutional Networks)

AlexNet의 구조에서 Dense layer 부분을 Convolution layer로 변형한 모델 (convolutionalization)

Semantic segmentation을 하기 위해서는 AlexNet 구조에서 FC layer를 변형 해야 하는 이유가 있다

1. FC layer를 사용하면 이미지의 위치 정보가 사라진다
2. FC layer를 사용하면 입력 이미지의 크기가 고정된다

따라서 semantic segmenation을 하기 위해서 마지막 FC layer를 convolution layer로
변형 함으로써 위치 정보를 보존하고 이미지 크기로 부터 유연할 수 있다

Skip combining

이전의 정보를 더함으로써 복원할때 활용한다
이전의 layer는 현재의 layer가 가지는 정보보다 디테일한 정보를 가지고 있기 때문에 복원하는데 도움이 되기 때문이다

U-net

Instance segmentation 방법

U-net 구조

U-Net은 이미지의 전반적인 컨텍스트 정보를 얻기 위한 네트워크와 정확한 지역화(Localization)를 위한 네트워크가 대칭 형태로 구성되어 있다

Contracting path의 최종 feature map으로 부터 높은 해상도의 결과를 얻기 위해 upsampling을 진행한다
coarse map -> dense prediction

Overlap-Tile Input

이미지를 타일로 나누어서 입력으로 사용한다
파란 영역의 이미지를 입력하면 노란 영역의 segmentation 결과를 얻는다

다음 tile에 대한 segmentation을 얻기 위해서는 이전 입력의 일부분이 포함되어 있어야 한다

이미지의 경계 부분 픽셀에 대한 segmentation을 위해 0이나 임의의 패딩값을 사용하는 대신
이미지 경계 부분의 미러링을 이용한 Extrapolation 기법을 사용한다

U-net 구현하기

import tensorflow as tf 
import tensorflow_datasets as tfds
import matplotlib.pyplot as plt

dataset, info = tfds.load('oxford_iiit_pet:3.*.*', with_info=True) # 37마리 애완동물 6000장 이미지 

dataset['train'] # cpu, gpu 둘다 활용하여 효율적 연산 가능한 데이터 구조 
# <PrefetchDataset shapes: {file_name: (), image: (None, None, 3), label: (), segmentation_mask: (None, None, 1), species: ()}, types: {file_name: tf.string, image: tf.uint8, label: tf.int64, segmentation_mask: tf.uint8, species: tf.int64}>

dataset['test']

# <PrefetchDataset shapes: {file_name: (), image: (None, None, 3), label: (), segmentation_mask: (None, None, 1), species: ()}, types: {file_name: tf.string, image: tf.uint8, label: tf.int64, segmentation_mask: tf.uint8, species: tf.int64}>

for i in dataset['train'].take(1):
    image = i['image']
    mask = i['segmentation_mask']

fig, ax = plt.subplots(1,2)
ax[0].imshow(image)
ax[1].imshow(mask, cmap='gray')

backbone = tf.keras.applications.MobileNetV2(input_shape=(128,128,3), include_top=False)

backbone.summary()
Model: "mobilenetv2_1.00_128"
__________________________________________________________________________________________________
Layer (type)                    Output Shape         Param #     Connected to                     
==================================================================================================
input_1 (InputLayer)            [(None, 128, 128, 3) 0                                            
__________________________________________________________________________________________________
Conv1 (Conv2D)                  (None, 64, 64, 32)   864         input_1[0][0]                    
__________________________________________________________________________________________________
bn_Conv1 (BatchNormalization)   (None, 64, 64, 32)   128         Conv1[0][0]                      
__________________________________________________________________________________________________
Conv1_relu (ReLU)               (None, 64, 64, 32)   0           bn_Conv1[0][0]                   
__________________________________________________________________________________________________
expanded_conv_depthwise (Depthw (None, 64, 64, 32)   288         Conv1_relu[0][0]                 
__________________________________________________________________________________________________
expanded_conv_depthwise_BN (Bat (None, 64, 64, 32)   128         expanded_conv_depthwise[0][0]    
__________________________________________________________________________________________________
expanded_conv_depthwise_relu (R (None, 64, 64, 32)   0           expanded_conv_depthwise_BN[0][0] 
__________________________________________________________________________________________________
expanded_conv_project (Conv2D)  (None, 64, 64, 16)   512         expanded_conv_depthwise_relu[0][0
__________________________________________________________________________________________________
expanded_conv_project_BN (Batch (None, 64, 64, 16)   64          expanded_conv_project[0][0]      
__________________________________________________________________________________________________
block_1_expand (Conv2D)         (None, 64, 64, 96)   1536        expanded_conv_project_BN[0][0]   
__________________________________________________________________________________________________
block_1_expand_BN (BatchNormali (None, 64, 64, 96)   384         block_1_expand[0][0]             
__________________________________________________________________________________________________
block_1_expand_relu (ReLU)      (None, 64, 64, 96)   0           block_1_expand_BN[0][0]          
__________________________________________________________________________________________________
block_1_pad (ZeroPadding2D)     (None, 65, 65, 96)   0           block_1_expand_relu[0][0]        
__________________________________________________________________________________________________
block_1_depthwise (DepthwiseCon (None, 32, 32, 96)   864         block_1_pad[0][0]                
__________________________________________________________________________________________________
block_1_depthwise_BN (BatchNorm (None, 32, 32, 96)   384         block_1_depthwise[0][0]          
__________________________________________________________________________________________________
block_1_depthwise_relu (ReLU)   (None, 32, 32, 96)   0           block_1_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_1_project (Conv2D)        (None, 32, 32, 24)   2304        block_1_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_1_project_BN (BatchNormal (None, 32, 32, 24)   96          block_1_project[0][0]            
__________________________________________________________________________________________________
block_2_expand (Conv2D)         (None, 32, 32, 144)  3456        block_1_project_BN[0][0]         
__________________________________________________________________________________________________
block_2_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_2_expand[0][0]             
__________________________________________________________________________________________________
block_2_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_2_expand_BN[0][0]          
__________________________________________________________________________________________________
block_2_depthwise (DepthwiseCon (None, 32, 32, 144)  1296        block_2_expand_relu[0][0]        
__________________________________________________________________________________________________
block_2_depthwise_BN (BatchNorm (None, 32, 32, 144)  576         block_2_depthwise[0][0]          
__________________________________________________________________________________________________
block_2_depthwise_relu (ReLU)   (None, 32, 32, 144)  0           block_2_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_2_project (Conv2D)        (None, 32, 32, 24)   3456        block_2_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_2_project_BN (BatchNormal (None, 32, 32, 24)   96          block_2_project[0][0]            
__________________________________________________________________________________________________
block_2_add (Add)               (None, 32, 32, 24)   0           block_1_project_BN[0][0]         
                                                                 block_2_project_BN[0][0]         
__________________________________________________________________________________________________
block_3_expand (Conv2D)         (None, 32, 32, 144)  3456        block_2_add[0][0]                
__________________________________________________________________________________________________
block_3_expand_BN (BatchNormali (None, 32, 32, 144)  576         block_3_expand[0][0]             
__________________________________________________________________________________________________
block_3_expand_relu (ReLU)      (None, 32, 32, 144)  0           block_3_expand_BN[0][0]          
__________________________________________________________________________________________________
block_3_pad (ZeroPadding2D)     (None, 33, 33, 144)  0           block_3_expand_relu[0][0]        
__________________________________________________________________________________________________
block_3_depthwise (DepthwiseCon (None, 16, 16, 144)  1296        block_3_pad[0][0]                
__________________________________________________________________________________________________
block_3_depthwise_BN (BatchNorm (None, 16, 16, 144)  576         block_3_depthwise[0][0]          
__________________________________________________________________________________________________
block_3_depthwise_relu (ReLU)   (None, 16, 16, 144)  0           block_3_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_3_project (Conv2D)        (None, 16, 16, 32)   4608        block_3_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_3_project_BN (BatchNormal (None, 16, 16, 32)   128         block_3_project[0][0]            
__________________________________________________________________________________________________
block_4_expand (Conv2D)         (None, 16, 16, 192)  6144        block_3_project_BN[0][0]         
__________________________________________________________________________________________________
block_4_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_4_expand[0][0]             
__________________________________________________________________________________________________
block_4_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_4_expand_BN[0][0]          
__________________________________________________________________________________________________
block_4_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_4_expand_relu[0][0]        
__________________________________________________________________________________________________
block_4_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_4_depthwise[0][0]          
__________________________________________________________________________________________________
block_4_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_4_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_4_project (Conv2D)        (None, 16, 16, 32)   6144        block_4_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_4_project_BN (BatchNormal (None, 16, 16, 32)   128         block_4_project[0][0]            
__________________________________________________________________________________________________
block_4_add (Add)               (None, 16, 16, 32)   0           block_3_project_BN[0][0]         
                                                                 block_4_project_BN[0][0]         
__________________________________________________________________________________________________
block_5_expand (Conv2D)         (None, 16, 16, 192)  6144        block_4_add[0][0]                
__________________________________________________________________________________________________
block_5_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_5_expand[0][0]             
__________________________________________________________________________________________________
block_5_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_5_expand_BN[0][0]          
__________________________________________________________________________________________________
block_5_depthwise (DepthwiseCon (None, 16, 16, 192)  1728        block_5_expand_relu[0][0]        
__________________________________________________________________________________________________
block_5_depthwise_BN (BatchNorm (None, 16, 16, 192)  768         block_5_depthwise[0][0]          
__________________________________________________________________________________________________
block_5_depthwise_relu (ReLU)   (None, 16, 16, 192)  0           block_5_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_5_project (Conv2D)        (None, 16, 16, 32)   6144        block_5_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_5_project_BN (BatchNormal (None, 16, 16, 32)   128         block_5_project[0][0]            
__________________________________________________________________________________________________
block_5_add (Add)               (None, 16, 16, 32)   0           block_4_add[0][0]                
                                                                 block_5_project_BN[0][0]         
__________________________________________________________________________________________________
block_6_expand (Conv2D)         (None, 16, 16, 192)  6144        block_5_add[0][0]                
__________________________________________________________________________________________________
block_6_expand_BN (BatchNormali (None, 16, 16, 192)  768         block_6_expand[0][0]             
__________________________________________________________________________________________________
block_6_expand_relu (ReLU)      (None, 16, 16, 192)  0           block_6_expand_BN[0][0]          
__________________________________________________________________________________________________
block_6_pad (ZeroPadding2D)     (None, 17, 17, 192)  0           block_6_expand_relu[0][0]        
__________________________________________________________________________________________________
block_6_depthwise (DepthwiseCon (None, 8, 8, 192)    1728        block_6_pad[0][0]                
__________________________________________________________________________________________________
block_6_depthwise_BN (BatchNorm (None, 8, 8, 192)    768         block_6_depthwise[0][0]          
__________________________________________________________________________________________________
block_6_depthwise_relu (ReLU)   (None, 8, 8, 192)    0           block_6_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_6_project (Conv2D)        (None, 8, 8, 64)     12288       block_6_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_6_project_BN (BatchNormal (None, 8, 8, 64)     256         block_6_project[0][0]            
__________________________________________________________________________________________________
block_7_expand (Conv2D)         (None, 8, 8, 384)    24576       block_6_project_BN[0][0]         
__________________________________________________________________________________________________
block_7_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_7_expand[0][0]             
__________________________________________________________________________________________________
block_7_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_7_expand_BN[0][0]          
__________________________________________________________________________________________________
block_7_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_7_expand_relu[0][0]        
__________________________________________________________________________________________________
block_7_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_7_depthwise[0][0]          
__________________________________________________________________________________________________
block_7_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_7_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_7_project (Conv2D)        (None, 8, 8, 64)     24576       block_7_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_7_project_BN (BatchNormal (None, 8, 8, 64)     256         block_7_project[0][0]            
__________________________________________________________________________________________________
block_7_add (Add)               (None, 8, 8, 64)     0           block_6_project_BN[0][0]         
                                                                 block_7_project_BN[0][0]         
__________________________________________________________________________________________________
block_8_expand (Conv2D)         (None, 8, 8, 384)    24576       block_7_add[0][0]                
__________________________________________________________________________________________________
block_8_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_8_expand[0][0]             
__________________________________________________________________________________________________
block_8_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_8_expand_BN[0][0]          
__________________________________________________________________________________________________
block_8_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_8_expand_relu[0][0]        
__________________________________________________________________________________________________
block_8_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_8_depthwise[0][0]          
__________________________________________________________________________________________________
block_8_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_8_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_8_project (Conv2D)        (None, 8, 8, 64)     24576       block_8_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_8_project_BN (BatchNormal (None, 8, 8, 64)     256         block_8_project[0][0]            
__________________________________________________________________________________________________
block_8_add (Add)               (None, 8, 8, 64)     0           block_7_add[0][0]                
                                                                 block_8_project_BN[0][0]         
__________________________________________________________________________________________________
block_9_expand (Conv2D)         (None, 8, 8, 384)    24576       block_8_add[0][0]                
__________________________________________________________________________________________________
block_9_expand_BN (BatchNormali (None, 8, 8, 384)    1536        block_9_expand[0][0]             
__________________________________________________________________________________________________
block_9_expand_relu (ReLU)      (None, 8, 8, 384)    0           block_9_expand_BN[0][0]          
__________________________________________________________________________________________________
block_9_depthwise (DepthwiseCon (None, 8, 8, 384)    3456        block_9_expand_relu[0][0]        
__________________________________________________________________________________________________
block_9_depthwise_BN (BatchNorm (None, 8, 8, 384)    1536        block_9_depthwise[0][0]          
__________________________________________________________________________________________________
block_9_depthwise_relu (ReLU)   (None, 8, 8, 384)    0           block_9_depthwise_BN[0][0]       
__________________________________________________________________________________________________
block_9_project (Conv2D)        (None, 8, 8, 64)     24576       block_9_depthwise_relu[0][0]     
__________________________________________________________________________________________________
block_9_project_BN (BatchNormal (None, 8, 8, 64)     256         block_9_project[0][0]            
__________________________________________________________________________________________________
block_9_add (Add)               (None, 8, 8, 64)     0           block_8_add[0][0]                
                                                                 block_9_project_BN[0][0]         
__________________________________________________________________________________________________
block_10_expand (Conv2D)        (None, 8, 8, 384)    24576       block_9_add[0][0]                
__________________________________________________________________________________________________
block_10_expand_BN (BatchNormal (None, 8, 8, 384)    1536        block_10_expand[0][0]            
__________________________________________________________________________________________________
block_10_expand_relu (ReLU)     (None, 8, 8, 384)    0           block_10_expand_BN[0][0]         
__________________________________________________________________________________________________
block_10_depthwise (DepthwiseCo (None, 8, 8, 384)    3456        block_10_expand_relu[0][0]       
__________________________________________________________________________________________________
block_10_depthwise_BN (BatchNor (None, 8, 8, 384)    1536        block_10_depthwise[0][0]         
__________________________________________________________________________________________________
block_10_depthwise_relu (ReLU)  (None, 8, 8, 384)    0           block_10_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_10_project (Conv2D)       (None, 8, 8, 96)     36864       block_10_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_10_project_BN (BatchNorma (None, 8, 8, 96)     384         block_10_project[0][0]           
__________________________________________________________________________________________________
block_11_expand (Conv2D)        (None, 8, 8, 576)    55296       block_10_project_BN[0][0]        
__________________________________________________________________________________________________
block_11_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_11_expand[0][0]            
__________________________________________________________________________________________________
block_11_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_11_expand_BN[0][0]         
__________________________________________________________________________________________________
block_11_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_11_expand_relu[0][0]       
__________________________________________________________________________________________________
block_11_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_11_depthwise[0][0]         
__________________________________________________________________________________________________
block_11_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_11_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_11_project (Conv2D)       (None, 8, 8, 96)     55296       block_11_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_11_project_BN (BatchNorma (None, 8, 8, 96)     384         block_11_project[0][0]           
__________________________________________________________________________________________________
block_11_add (Add)              (None, 8, 8, 96)     0           block_10_project_BN[0][0]        
                                                                 block_11_project_BN[0][0]        
__________________________________________________________________________________________________
block_12_expand (Conv2D)        (None, 8, 8, 576)    55296       block_11_add[0][0]               
__________________________________________________________________________________________________
block_12_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_12_expand[0][0]            
__________________________________________________________________________________________________
block_12_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_12_expand_BN[0][0]         
__________________________________________________________________________________________________
block_12_depthwise (DepthwiseCo (None, 8, 8, 576)    5184        block_12_expand_relu[0][0]       
__________________________________________________________________________________________________
block_12_depthwise_BN (BatchNor (None, 8, 8, 576)    2304        block_12_depthwise[0][0]         
__________________________________________________________________________________________________
block_12_depthwise_relu (ReLU)  (None, 8, 8, 576)    0           block_12_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_12_project (Conv2D)       (None, 8, 8, 96)     55296       block_12_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_12_project_BN (BatchNorma (None, 8, 8, 96)     384         block_12_project[0][0]           
__________________________________________________________________________________________________
block_12_add (Add)              (None, 8, 8, 96)     0           block_11_add[0][0]               
                                                                 block_12_project_BN[0][0]        
__________________________________________________________________________________________________
block_13_expand (Conv2D)        (None, 8, 8, 576)    55296       block_12_add[0][0]               
__________________________________________________________________________________________________
block_13_expand_BN (BatchNormal (None, 8, 8, 576)    2304        block_13_expand[0][0]            
__________________________________________________________________________________________________
block_13_expand_relu (ReLU)     (None, 8, 8, 576)    0           block_13_expand_BN[0][0]         
__________________________________________________________________________________________________
block_13_pad (ZeroPadding2D)    (None, 9, 9, 576)    0           block_13_expand_relu[0][0]       
__________________________________________________________________________________________________
block_13_depthwise (DepthwiseCo (None, 4, 4, 576)    5184        block_13_pad[0][0]               
__________________________________________________________________________________________________
block_13_depthwise_BN (BatchNor (None, 4, 4, 576)    2304        block_13_depthwise[0][0]         
__________________________________________________________________________________________________
block_13_depthwise_relu (ReLU)  (None, 4, 4, 576)    0           block_13_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_13_project (Conv2D)       (None, 4, 4, 160)    92160       block_13_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_13_project_BN (BatchNorma (None, 4, 4, 160)    640         block_13_project[0][0]           
__________________________________________________________________________________________________
block_14_expand (Conv2D)        (None, 4, 4, 960)    153600      block_13_project_BN[0][0]        
__________________________________________________________________________________________________
block_14_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_14_expand[0][0]            
__________________________________________________________________________________________________
block_14_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_14_expand_BN[0][0]         
__________________________________________________________________________________________________
block_14_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_14_expand_relu[0][0]       
__________________________________________________________________________________________________
block_14_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_14_depthwise[0][0]         
__________________________________________________________________________________________________
block_14_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_14_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_14_project (Conv2D)       (None, 4, 4, 160)    153600      block_14_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_14_project_BN (BatchNorma (None, 4, 4, 160)    640         block_14_project[0][0]           
__________________________________________________________________________________________________
block_14_add (Add)              (None, 4, 4, 160)    0           block_13_project_BN[0][0]        
                                                                 block_14_project_BN[0][0]        
__________________________________________________________________________________________________
block_15_expand (Conv2D)        (None, 4, 4, 960)    153600      block_14_add[0][0]               
__________________________________________________________________________________________________
block_15_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_15_expand[0][0]            
__________________________________________________________________________________________________
block_15_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_15_expand_BN[0][0]         
__________________________________________________________________________________________________
block_15_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_15_expand_relu[0][0]       
__________________________________________________________________________________________________
block_15_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_15_depthwise[0][0]         
__________________________________________________________________________________________________
block_15_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_15_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_15_project (Conv2D)       (None, 4, 4, 160)    153600      block_15_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_15_project_BN (BatchNorma (None, 4, 4, 160)    640         block_15_project[0][0]           
__________________________________________________________________________________________________
block_15_add (Add)              (None, 4, 4, 160)    0           block_14_add[0][0]               
                                                                 block_15_project_BN[0][0]        
__________________________________________________________________________________________________
block_16_expand (Conv2D)        (None, 4, 4, 960)    153600      block_15_add[0][0]               
__________________________________________________________________________________________________
block_16_expand_BN (BatchNormal (None, 4, 4, 960)    3840        block_16_expand[0][0]            
__________________________________________________________________________________________________
block_16_expand_relu (ReLU)     (None, 4, 4, 960)    0           block_16_expand_BN[0][0]         
__________________________________________________________________________________________________
block_16_depthwise (DepthwiseCo (None, 4, 4, 960)    8640        block_16_expand_relu[0][0]       
__________________________________________________________________________________________________
block_16_depthwise_BN (BatchNor (None, 4, 4, 960)    3840        block_16_depthwise[0][0]         
__________________________________________________________________________________________________
block_16_depthwise_relu (ReLU)  (None, 4, 4, 960)    0           block_16_depthwise_BN[0][0]      
__________________________________________________________________________________________________
block_16_project (Conv2D)       (None, 4, 4, 320)    307200      block_16_depthwise_relu[0][0]    
__________________________________________________________________________________________________
block_16_project_BN (BatchNorma (None, 4, 4, 320)    1280        block_16_project[0][0]           
__________________________________________________________________________________________________
Conv_1 (Conv2D)                 (None, 4, 4, 1280)   409600      block_16_project_BN[0][0]        
__________________________________________________________________________________________________
Conv_1_bn (BatchNormalization)  (None, 4, 4, 1280)   5120        Conv_1[0][0]                     
__________________________________________________________________________________________________
out_relu (ReLU)                 (None, 4, 4, 1280)   0           Conv_1_bn[0][0]                  
==================================================================================================
Total params: 2,257,984
Trainable params: 2,223,872
Non-trainable params: 34,112
__________________________________________________________________________________________________

from tensorflow.keras.utils import plot_model
plot_model(backbone)

len(backbone.layers)
# 154

layer_names = [
    'block_1_expand_relu', # 64x64
    'block_3_expand_relu', # 32x32
    'block_6_expand_relu', # 16x16
    'block_13_expand_relu', # 8x8
    'block_16_project', # 4x4
]
base_model_outputs = [base_model.get_layer(name).output for name in layer_names]

저작자표시 비영리

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

39일차 - 논문 수업 (GAN) (0)	2021.11.16
38일차 - Style Transfer (0)	2021.11.16
36일차 - 논문 수업 (Segmentation) (0)	2021.11.16
35일차 - 논문 수업 (Segmentation) (0)	2021.11.01
34일차 - 논문 수업 (VAE) (0)	2021.11.01

36일차 - 논문 수업 (Segmentation)

2021. 11. 16. 18:41

728x90

Variational AutoEncoder

Reparametization trick

확률의 문제를 학습의 문제로 변형하기 위한 방법
미분 가능한 식으로 변형하여 backpropagation이 가능하도록 변형한다

import tensorflow as tf 
from tensorflow.keras import backend as K # backend(tensorflow)에 맞추어 low level 데이터 연산을 해주는 라이브러리 
import inspect

print(inspect.getsource(K.random_normal)) # 정규분포로 부터 무작위 샘플을 추출 하는 함수 


@keras_export('keras.backend.random_normal')
@tf.__internal__.dispatch.add_dispatch_support
@doc_controls.do_not_generate_docs
def random_normal(shape, mean=0.0, stddev=1.0, dtype=None, seed=None):
  """Returns a tensor with normal distribution of values.

  It is an alias to `tf.random.normal`.

  Args:
      shape: A tuple of integers, the shape of tensor to create.
      mean: A float, the mean value of the normal distribution to draw samples.
        Default to 0.0.
      stddev: A float, the standard deviation of the normal distribution
        to draw samples. Default to 1.0.
      dtype: `tf.dtypes.DType`, dtype of returned tensor. Default to use Keras
        backend dtype which is float32.
      seed: Integer, random seed. Will use a random numpy integer when not
        specified.

  Returns:
      A tensor with normal distribution of values.

  Example:

  >>> random_normal_tensor = tf.keras.backend.random_normal(shape=(2,3),
  ... mean=0.0, stddev=1.0)
  >>> random_normal_tensor
  <tf.Tensor: shape=(2, 3), dtype=float32, numpy=...,
  dtype=float32)>
  """
  if dtype is None:
    dtype = floatx()
  if seed is None:
    seed = np.random.randint(10e6)
  return tf.random.normal(
      shape, mean=mean, stddev=stddev, dtype=dtype, seed=seed)

tf.keras.losses.kl_divergence

# <function keras.losses.kl_divergence(y_true, y_pred)>

K.random_normal((2,3))

# <tf.Tensor: shape=(2, 3), dtype=float32, numpy=
# array([[-0.03670657,  0.19654608,  1.8087223 ],
#        [-0.5491878 , -1.7109    ,  0.6098757 ]], dtype=float32)>

epsilon = K.random_normal(shape=(K.shape(z_mean)[0], latent_dim), mean=0, stddev=0.1)

random normal 자체는 미분이 가능하지만 random normal 결과는 미분가능하지 않기 때문에 파라미터화 해야 한다

from sklearn.linear_model import LinearRegression, Ridge, Lasso, ElasticNet

Regularizer term

Regularizer term은 λ또는 Regularization Parameter라고도 부른다
Regularizer term은 급격한 변화를 방지하는 역할을 한다. 결국 일반화 된 모델이 만들어지도록 유도한다

Lasso
- L1-norm 패널티를 가진 선형 회귀 방법, overfitting을 방지하는 방법이다
Ridge
- L2-norm 패널티를 가진 선형 회귀 방법, overfitting을 방지하는 방법이다
ElasticNet L1, L2
- L1-norm, L2-norm 패널티를 동시에 가진 선형 회귀 방법, overfitting을 방지하는 방법이다

Lasso

MSE를 최소로 만드는 w,b 찾기

MSE + penalty를 최소로 만드는 w, b 찾기

MSE와 penalty를 최소가 되는 w와 b를 찾는 것이 핵심이다
이는 w의 모든 원소가 0이되거나 0에 가깝게 되야한다는 의미이다
그렇다면 어떤 특성들은 모델을 만들때 사용하지 않게 된다
모델을 만들때 모든 가중치를 고려하게 된다면 복잡한 모델이 만들어지고 그만큼 학습 데이터에만 잘 학습된
즉, overfitting이 된 모델이 탄생하게 될 것이다

따라서 penalty를 추가함으로써 자연스럽게 학습과정에서 덜 중요한 특징들은 사용되지 않게 적절한 특징을
추려내는 것이 관건이다
예를 들어 100개 특성을 갖는 회귀 모델일 있다고 가정할 때 α=1로 설정했더니 95개가 0이 되면서 5개의 특성이 사용된 것을 확인 했다. 이때 성능 측정을 했더니 underfitting이 발생했다
그런데 α=0.0001로 설정했더니 5개만 0이 되면서 95개 특성을 사용해야 했다. 이때 성능 측정을 했더니 overfitting이 발생했다
결국 α=0.1로 설정하여 70개 특성이 0이 되면서 30개 특성을 사용하여 성능 측정을 했더니 적절하게 일반화 된 모델을 만들 수 있게 되었다
이때 이 α를 구할 수 있도록 파라미터가 추가 된 것은 L1-norm penalty를 추가했기 때문에 가능한 일이다

Ridge

Ridge는 L1-norm 대신 L2-norm을 사용한 Regulaization 방법이라고 생각하면 된다
Lasso와 차이점이 있다면 가중치가 0에 가까워질 뿐 0이 되지는 않는다
왜냐 하면 Ridge는 절대값을 추가한 것이 아니라 제곱합이기 때문에 0이 되지 않는다

따라서 특성의 중요도가 전체적으로 비슷한 경우 Ridge가 좀 더 괜찮은 모델을 찾아줄 것이다

layer = tf.keras.layers.Dense(
    5, input_dim=5,
    kernel_initializer='ones',
    kernel_regularizer=tf.keras.regularizers.L1(0.01)
)

layer.weights
# []

layer.build((None,5))

layer.weights

[<tf.Variable 'kernel:0' shape=(5, 5) dtype=float32, numpy=
 array([[1., 1., 1., 1., 1.],
        [1., 1., 1., 1., 1.],
        [1., 1., 1., 1., 1.],
        [1., 1., 1., 1., 1.],
        [1., 1., 1., 1., 1.]], dtype=float32)>,
 <tf.Variable 'bias:0' shape=(5,) dtype=float32, numpy=array([0., 0., 0., 0., 0.], dtype=float32)>]

tensor = tf.ones(shape=(5,5))*2.0

tensor
<tf.Tensor: shape=(5, 5), dtype=float32, numpy=
array([[2., 2., 2., 2., 2.],
       [2., 2., 2., 2., 2.],
       [2., 2., 2., 2., 2.],
       [2., 2., 2., 2., 2.],
       [2., 2., 2., 2., 2.]], dtype=float32)>

out = layer(tensor)
out
<tf.Tensor: shape=(5, 5), dtype=float32, numpy=
array([[10., 10., 10., 10., 10.],
       [10., 10., 10., 10., 10.],
       [10., 10., 10., 10., 10.],
       [10., 10., 10., 10., 10.],
       [10., 10., 10., 10., 10.]], dtype=float32)>

Regularizer / penalty

각 항의 계수를 작게 만들어 가설함수를 간단하게 하고 과적합을 피할수 있도록 유도한다

loss = loss + regulizer_term(L1, L2)

(X_train, y_train), (X_test, y_test) = tf.keras.datasets.mnist.load_data()
y_train = tf.cast(y_train, tf.int32)

model = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(input_shape=(28,28)),
    tf.keras.layers.Dense(128,activation='relu', kernel_regularizer=tf.keras.regularizers.l1()),
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Dense(10, activation='softmax')
])

loss = tf.keras.losses.SparseCategoricalCrossentropy()

prediction = model(X_train)

model.losses # regularizer를 반환한다
# [<tf.Tensor: shape=(), dtype=float32, numpy=40.870396>]

# loss + Regularization term / overfitting은 덜 나지만 학습속도는 느려진다
loss(y_train, prediction) + tf.reduce_sum(model.losses) 
# <tf.Tensor: shape=(), dtype=float32, numpy=164.54881>

loss function의 차이

(X_train, y_train), (X_test, y_test) = tf.keras.datasets.mnist.load_data()
y_train = tf.cast(y_train, tf.int32)
y_train_ohe = tf.keras.utils.to_categorical(y_train)

model1 = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(input_shape=(28,28)),
    tf.keras.layers.Dense(128,activation='relu', kernel_regularizer=tf.keras.regularizers.l1()),
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Dense(10, activation='softmax')
])
prediction = model1(X_train)

tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
# tf.keras.losses.SparseCategoricalCrossentropy()(y_train, prediction)
# tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(60000,), dtype=float32, numpy=
# array([192.52524 ,  51.914566,  72.927284, ..., 171.85602 , 150.09734 ,
#         53.307495], dtype=float32)>

# tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
tf.keras.losses.SparseCategoricalCrossentropy()(y_train, prediction)
# tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(), dtype=float32, numpy=116.44715>

# tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
# tf.keras.losses.SparseCategoricalCrossentropy()(y_train, prediction)
tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(60000,), dtype=float32, numpy=
# array([2.46115  , 2.46115  , 2.46115  , ..., 2.46115  , 2.46115  ,
#        2.4379332], dtype=float32)>

model2 = tf.keras.models.Sequential([
    tf.keras.layers.Flatten(input_shape=(28,28)),
    tf.keras.layers.Dense(128,activation='relu', kernel_regularizer=tf.keras.regularizers.l1()),
    tf.keras.layers.Dropout(0.2),
    tf.keras.layers.Dense(10)
])
prediction = model2(X_train)

tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
# tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)(y_train, prediction)
# tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(60000,), dtype=float32, numpy=
# array([17.216707 , 17.727533 ,  1.6094378, ...,  1.0986124,  1.0986124,
#        17.216707 ], dtype=float32)>

# tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)(y_train, prediction)
# tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(), dtype=float32, numpy=151.73372>

# tf.keras.losses.sparse_categorical_crossentropy(y_train, prediction)
# tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True)(y_train, prediction)
tf.nn.sparse_softmax_cross_entropy_with_logits(y_train, prediction)

# <tf.Tensor: shape=(60000,), dtype=float32, numpy=
# array([114.73    , 230.42961 , 114.96562 , ...,  13.309306,  -0.      ,
#        137.89368 ], dtype=float32)>

# model1.compile(loss=tf.keras.losses.sparse_categorical_crossentropy)
# model1.compile(loss=tf.keras.losses.SparseCategoricalCrossentropy())
model1.compile(loss=tf.nn.sparse_softmax_cross_entropy_with_logits)
# model1.compile(loss=tf.nn.softmax_cross_entropy_with_logits)

# model2.compile(loss=tf.keras.losses.sparse_categorical_crossentropy)
# model2.compile(loss=tf.keras.losses.SparseCategoricalCrossentropy())
model2.compile(loss=tf.nn.sparse_softmax_cross_entropy_with_logits)
# model2.compile(loss=tf.nn.softmax_cross_entropy_with_logits)

model1.fit(X_train, y_train, epochs=2) # sparse_categorical_crossentropy

Epoch 1/2
1875/1875 [==============================] - 6s 3ms/step - loss: 9.0639
Epoch 2/2
1875/1875 [==============================] - 6s 3ms/step - loss: 3.7299
<keras.callbacks.History at 0x2ab863250>

model1.fit(X_train, y_train, epochs=2) # SparseCategoricalCrossentropy

Epoch 1/2
32/1875 [..............................] - ETA: 6s - loss: 6.2577 
1875/1875 [==============================] - 6s 3ms/step - loss: 3.1716
Epoch 2/2
1875/1875 [==============================] - 6s 3ms/step - loss: 2.5536
<keras.callbacks.History at 0x2afab4bb0>

model1.fit(X_train, y_train_ohe, epochs=2) # nn.sparse_softmax_cross_entropy_with_logits

model1.fit(X_train, y_train, epochs=2) # nn.softmax_cross_entropy_with_logits

Epoch 1/2
  18/1875 [..............................] - ETA: 12s - loss: 97.9138 
  1875/1875 [==============================] - 9s 5ms/step - loss: 103.1872
Epoch 2/2
1875/1875 [==============================] - 9s 5ms/step - loss: 103.0049
<keras.callbacks.History at 0x2abb74610>

model2.fit(X_train, y_train_ohe, epochs=2) # nn.softmax_cross_entropy_with_logits (with softmax)

Epoch 1/2
  21/1875 [..............................] - ETA: 9s - loss: 91.7908
  1875/1875 [==============================] - 9s 5ms/step - loss: 8.9059
Epoch 2/2
1875/1875 [==============================] - 8s 4ms/step - loss: 3.7517
<keras.callbacks.History at 0x2ca7eba00>

model2.fit(X_train, y_train_ohe, epochs=2) # nn.sparse_softmax_cross_entropy_with_logits

One-hot encoding x => sparse O
One-hot encoding O => Sparse X

(X_train, y_train), (X_test, y_test) = tf.keras.datasets.mnist.load_data()

model = tf.keras.models.Sequential([
    tf.keras.layers.Conv2D(3,3),
    tf.keras.layers.Conv2D(1,3)
])

model(X_train, apply_sigmoid=True)

tensor = tf.ones(shape)

out = layer(tensor, apply_sigmoid=True)

저작자표시 비영리

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

38일차 - Style Transfer (0)	2021.11.16
37일차 - 논문 수업 (Segmentation) (0)	2021.11.16
35일차 - 논문 수업 (Segmentation) (0)	2021.11.01
34일차 - 논문 수업 (VAE) (0)	2021.11.01
33일차 - 논문 수업 (VAE) (0)	2021.11.01

18. 세계음주 데이터 분석

2021. 11. 3. 00:30

728x90

drinks.csv

0.00MB

import pandas as pd
drinks = pd.read_csv('drinks.csv')
drinks.info

<bound method DataFrame.info of          country  beer_servings  spirit_servings  wine_servings  \
0    Afghanistan              0                0              0   
1        Albania             89              132             54   
2        Algeria             25                0             14   
3        Andorra            245              138            312   
4         Angola            217               57             45   
..           ...            ...              ...            ...   
188    Venezuela            333              100              3   
189      Vietnam            111                2              1   
190        Yemen              6                0              0   
191       Zambia             32               19              4   
192     Zimbabwe             64               18              4   

     total_litres_of_pure_alcohol continent  
0                             0.0        AS  
1                             4.9        EU  
2                             0.7        AF  
3                            12.4        EU  
4                             5.9        AF  
..                            ...       ...  
188                           7.7        SA  
189                           2.0        AS  
190                           0.1        AS  
191                           2.5        AF  
192                           4.7        AF  

[193 rows x 6 columns]>

drinks.head

<bound method NDFrame.head of          country  beer_servings  spirit_servings  wine_servings  \
0    Afghanistan              0                0              0   
1        Albania             89              132             54   
2        Algeria             25                0             14   
3        Andorra            245              138            312   
4         Angola            217               57             45   
..           ...            ...              ...            ...   
188    Venezuela            333              100              3   
189      Vietnam            111                2              1   
190        Yemen              6                0              0   
191       Zambia             32               19              4   
192     Zimbabwe             64               18              4   

     total_litres_of_pure_alcohol continent  
0                             0.0        AS  
1                             4.9        EU  
2                             0.7        AF  
3                            12.4        EU  
4                             5.9        AF  
..                            ...       ...  
188                           7.7        SA  
189                           2.0        AS  
190                           0.1        AS  
191                           2.5        AF  
192                           4.7        AF  

[193 rows x 6 columns]>

# 피처 상관관계
# 피어스 상관계수 
# 'beer_serving', 'wine_servings'
corr = drinks[['beer_servings', 'wine_servings']].corr(method = 'pearson')
corr


	beer_servings	wine_servings
beer_servings	1.000000	0.527172
wine_servings	0.527172	1.000000

corr = drinks[['beer_servings','spirit_servings', 'wine_servings','total_litres_of_pure_alcohol']].corr(method = 'pearson')
corr

	beer_servings	spirit_servings	wine_servings	total_litres_of_pure_alcohol
beer_servings	1.000000	0.458819	0.527172	0.835839
spirit_servings	0.458819	1.000000	0.194797	0.654968
wine_servings	0.527172	0.194797	1.000000	0.667598
total_litres_of_pure_alcohol	0.835839	0.654968	0.667598	1.000000

# 상관계수 시각화
import matplotlib.pyplot as plt
import seaborn as sns
cols_view = ['beer','spirit', 'wine', 'alcohol']
sns.set(font_scale = 1.5)
hm = sns.heatmap(corr.values, cbar = True, annot = True, square=True,
                fmt = '.2f', annot_kws = {'size':15},
                yticklabels = cols_view, xticklabels = cols_view)
plt.show()

hm = sns.pairplot(drinks)
plt.show()

drinks.isnull().sum()
# drinks.info()

country                          0
beer_servings                    0
spirit_servings                  0
wine_servings                    0
total_litres_of_pure_alcohol     0
continent                       23
dtype: int64

drinks['continent'] = drinks['continent'].fillna('OT')

drinks['continent'].value_counts()

AF    53
EU    45
AS    44
OT    23
OC    16
SA    12
Name: continent, dtype: int64

# 대륙별 국가수 출력
print(drinks.groupby('continent').count()['country'])

continent
AF    53
AS    44
EU    45
OC    16
OT    23
SA    12
Name: country, dtype: int64

plt.pie(drinks['continent'].value_counts(),
        labels = drinks['continent'].value_counts().index.tolist(),
       autopct='%.0f%%',
       explode = (0,0,0,0.2,0,0),
       shadow=True)
plt.title('null data to "ot"')
plt.show()

drinks.groupby('continent')['spirit_servings'].max()

continent
AF    152
AS    326
EU    373
OC    254
OT    438
SA    302
Name: spirit_servings, dtype: int64

drinks.groupby('continent')['spirit_servings'].mean()

continent
AF     16.339623
AS     60.840909
EU    132.555556
OC     58.437500
OT    165.739130
SA    114.750000
Name: spirit_servings, dtype: float64

drinks.groupby('continent')['spirit_servings'].agg(['mean','min','max','sum'])

	mean	min	max	sum
continent				
AF	16.339623	0	152	866
AS	60.840909	0	326	2677
EU	132.555556	0	373	5965
OC	58.437500	0	254	935
OT	165.739130	68	438	3812
SA	114.750000	25	302	1377

dm = drinks['total_litres_of_pure_alcohol'].mean()
con_mean = drinks.groupby('continent')['total_litres_of_pure_alcohol'].mean()
con_mean[con_mean >= dm]

continent
EU    8.617778
OT    5.995652
SA    6.308333
Name: total_litres_of_pure_alcohol, dtype: float64

dmax = drinks.groupby('continent')['beer_servings'].mean()
dmax[dmax == dmax.max()]

continent
EU    193.777778
Name: beer_servings, dtype: float64

drinks.groupby('continent')['beer_servings'].mean().idxmax()

# 'EU'

drinks.groupby('continent')['beer_servings'].mean().idxmin()

# 'AS'

result

	mean	min	max	sum
continent				
AF	16.339623	0	152	866
AS	60.840909	0	326	2677
EU	132.555556	0	373	5965
OC	58.437500	0	254	935
OT	165.739130	68	438	3812
SA	114.750000	25	302	1377

result.index
# Index(['AF', 'AS', 'EU', 'OC', 'OT', 'SA'], dtype='object', name='continent')

import numpy as np
# result = drinks.groupby('continent')['beer_servings'].agg(['mean', 'min', 'max', 'sum']
means = result['mean'].tolist() 
mins = result['min'].tolist() 
maxs = result['max'].tolist()  
sums = result['sum'].tolist()                                                            
index = np.arange(len(result.index))
bar_width = 0.1
rects1 = plt.bar(index, means, bar_width, color = 'r', label = 'Mean')
rects2 = plt.bar(index, mins, bar_width, color = 'g', label = 'Min')
rects3 = plt.bar(index, maxs, bar_width, color = 'b', label = 'Max')
rects4 = plt.bar(index, sums, bar_width, color = 'y', label = 'Sum')
plt.xticks(index, result.index.tolist())
plt.legend(loc="best")
plt.show()

# 대륙별 total_litres_of_pure_alcohol 섭취량 평균을 시각화
import numpy as np
continent_mean = drinks.groupby('continent')['total_litres_of_pure_alcohol'].mean()
total_mean = drinks.total_litres_of_pure_alcohol.mean()

continents = continent_mean.index.tolist()
continents.append('Mean')

x_pos = np.arange(len(continents))
alcohol = continent_mean.tolist()
alcohol.append(total_mean)

bar_list = plt.bar(x_pos, alcohol, align = 'center', alpha = 0.5)
bar_list[len(continents)-1].set_color('r')
plt.plot([0., 6], [total_mean, total_mean], "k--")
plt.xticks(x_pos, continents)
plt.ylabel('total_litres_of_pure_alcohol')
plt.title('total_litres_of_pure_alcohol by continent')
plt.show()

# 대륙별 beer_serving 합계를 막대그래프로 시각화
# eu 막대의 색상을 빨강색으로 변경하기
# 전체 맥주 소비량 합계의 평균을 구해서 막대 그래프에 추가
# 평균선을 출력하기, 막대 색상은 노랑색
# 평균 선은 검정색("k--")
beer_sum = drinks.groupby('continent')['beer_servings'].sum()
beer_sum

continent
AF    3258
AS    1630
EU    8720
OC    1435
OT    3345
SA    2101
Name: beer_servings, dtype: int64

beer_mean = beer_sum.mean()
beer_mean
# 3414.8333333333335

continents = beer_sum.index.tolist()
continents.append("Mean")
continents

# ['AF', 'AS', 'EU', 'OC', 'OT', 'SA', 'Mean']

x_pos = np.arange(len(continents))
alcohol = beer_sum.tolist()
alcohol.append(beer_mean)
alcohol

[3258, 1630, 8720, 1435, 3345, 2101, 3414.8333333333335]

bar_list = plt.bar(x_pos, alcohol, align='center', alpha = 0.5)
bar_list[2].set_color("r")

저작자표시 비영리

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

20. 서울시 인구분석 \|\| 다중회귀 (0)	2021.11.23
19. 세계음주데이터2 (0)	2021.11.23
17. 서울 기온 분석 (0)	2021.11.02
16. EDA, 멕시코식당 주문 CHIPOTLE (0)	2021.10.28
15. 스크래핑 (0)	2021.10.28

17. 서울 기온 분석

2021. 11. 2. 23:38

728x90

seoul_5.csv

1.13MB

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns
df = pd.read_csv('seoul_5.csv', encoding = 'cp949')
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 40221 entries, 0 to 40220
Data columns (total 5 columns):
 #   Column   Non-Null Count  Dtype  
---  ------   --------------  -----  
 0   날짜       40221 non-null  object 
 1   지점       40221 non-null  int64  
 2   평균기온(℃)  39465 non-null  float64
 3   최저기온(℃)  39464 non-null  float64
 4   최고기온(℃)  39463 non-null  float64
dtypes: float64(3), int64(1), object(1)
memory usage: 1.5+ MB

df.describe()

	지점	평균기온(℃)	최저기온(℃)	최고기온(℃)
count	40221.0	39465.000000	39464.000000	39463.000000
mean	108.0	11.704019	7.406393	16.716083
std	0.0	10.668056	10.891154	10.998383
min	108.0	-19.200000	-23.100000	-16.300000
25%	108.0	2.600000	-1.500000	7.200000
50%	108.0	12.900000	8.000000	18.600000
75%	108.0	21.200000	17.000000	26.200000
max	108.0	33.700000	30.300000	39.600000

df

	날짜	지점	평균기온(℃)	최저기온(℃)	최고기온(℃)
0	1907-10-01	108	13.5	7.9	20.7
1	1907-10-02	108	16.2	7.9	22.0
2	1907-10-03	108	16.2	13.1	21.3
3	1907-10-04	108	16.5	11.2	22.0
4	1907-10-05	108	17.6	10.9	25.4
...	...	...	...	...	...
40216	2019-01-13	108	1.2	-3.0	7.6
40217	2019-01-14	108	1.4	-2.4	5.3
40218	2019-01-15	108	-1.7	-7.2	2.6
40219	2019-01-16	108	-5.2	-10.1	-1.1
40220	2019-01-17	108	-0.3	-3.2	4.0

# 결측치 제거
# 날짜 dateTime => 월일 잘라내기
df['날짜'] = pd.to_datetime(df['날짜'])
df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 40221 entries, 0 to 40220
Data columns (total 5 columns):
 #   Column   Non-Null Count  Dtype         
---  ------   --------------  -----         
 0   날짜       40221 non-null  datetime64[ns]
 1   지점       40221 non-null  int64         
 2   평균기온(℃)  39465 non-null  float64       
 3   최저기온(℃)  39464 non-null  float64       
 4   최고기온(℃)  39463 non-null  float64       
dtypes: datetime64[ns](1), float64(3), int64(1)
memory usage: 1.5 MB

df['Year'] = df['날짜'].dt.year
df['Month'] = df['날짜'].dt.month
df['Day'] = df['날짜'].dt.day
df.head()


날짜	지점	평균기온(℃)	최저기온(℃)	최고기온(℃)	Year	Month	Day
0	1907-10-01	108	13.5	7.9	20.7	1907	10	1
1	1907-10-02	108	16.2	7.9	22.0	1907	10	2
2	1907-10-03	108	16.2	13.1	21.3	1907	10	3
3	1907-10-04	108	16.5	11.2	22.0	1907	10	4
4	1907-10-05	108	17.6	10.9	25.4	1907	10	5

df.columns = ['날짜','지점','평균기온','최저기온','최고기온','Year','Month','Day']
df.head()

	날짜	지점	평균기온	최저기온	최고기온	Year	Month	Day
0	1907-10-01	108	13.5	7.9	20.7	1907	10	1
1	1907-10-02	108	16.2	7.9	22.0	1907	10	2
2	1907-10-03	108	16.2	13.1	21.3	1907	10	3
3	1907-10-04	108	16.5	11.2	22.0	1907	10	4
4	1907-10-05	108	17.6	10.9	25.4	1907	10	5

df_0214 = df[(df['Month'] == 2)&(df['Day'] == 14)]
df_0214

	날짜	지점	평균기온	최저기온	최고기온	Year	Month	Day
136	1908-02-14	108	-3.3	-7.5	2.3	1908	2	14
502	1909-02-14	108	2.6	-4.5	8.8	1909	2	14
867	1910-02-14	108	-3.1	-10.1	2.8	1910	2	14
1232	1911-02-14	108	0.8	0.0	3.5	1911	2	14
1597	1912-02-14	108	6.3	0.9	11.2	1912	2	14
...	...	...	...	...	...	...	...	...
38422	2014-02-14	108	2.7	-0.7	7.6	2014	2	14
38787	2015-02-14	108	2.0	-3.1	6.6	2015	2	14
39152	2016-02-14	108	-2.6	-6.8	5.8	2016	2	14
39518	2017-02-14	108	0.3	-4.0	6.5	2017	2	14
39883	2018-02-14	108	3.5	-0.7	8.7	2018	2	14
110 rows × 8 columns

df_year = df_0214['Year']
df_high = df_0214['최고기온']
df_low = df_0214['최저기온']

plt.style.use('ggplot')
fig = plt.figure(figsize=(20,5))
ax = fig.add_subplot(1,1,1) 

ax.plot(df_year, df_0214['최고기온'], label = df['날짜'])
ax.plot(df_year, df_0214['최저기온'])
ax.set_title('2월 14일 기온 변화')
ax.set_xticklabels(df['Year'], rotation = 90)
ax.set_xlabel('year')
ax.set_ylabel('temperture')
plt.show()

df.head()

	날짜	지점	평균기온	최저기온	최고기온	Year	Month	Day
0	1907-10-01	108	13.5	7.9	20.7	1907	10	1
1	1907-10-02	108	16.2	7.9	22.0	1907	10	2
2	1907-10-03	108	16.2	13.1	21.3	1907	10	3
3	1907-10-04	108	16.5	11.2	22.0	1907	10	4
4	1907-10-05	108	17.6	10.9	25.4	1907	10	5

plt.style.use('ggplot')
plt.hist(df['평균기온'], bins = 100, color = 'r')
plt.show()

df.isnull().sum()

날짜         0
지점         0
평균기온     756
최저기온     757
최고기온     758
Year       0
Month      0
Day        0
dtype: int64

df = df.dropna(axis=0, how='any', thresh=None, subset=None, inplace=False)
df.isnull().sum()

날짜       0
지점       0
평균기온     0
최저기온     0
최고기온     0
Year     0
Month    0
Day      0
dtype: int64

df.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 39463 entries, 0 to 40220
Data columns (total 8 columns):
 #   Column  Non-Null Count  Dtype         
---  ------  --------------  -----         
 0   날짜      39463 non-null  datetime64[ns]
 1   지점      39463 non-null  int64         
 2   평균기온    39463 non-null  float64       
 3   최저기온    39463 non-null  float64       
 4   최고기온    39463 non-null  float64       
 5   Year    39463 non-null  int64         
 6   Month   39463 non-null  int64         
 7   Day     39463 non-null  int64         
dtypes: datetime64[ns](1), float64(3), int64(4)
memory usage: 2.7 MB

plt.style.use('ggplot')
plt.boxplot(df['평균기온'])
plt.title("1907년부터 2019년까지 서울의 평균기온")

# Text(0.5, 1.0, '1907년부터 2019년까지 서울의 평균기온')

저작자표시 비영리

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

19. 세계음주데이터2 (0)	2021.11.23
18. 세계음주 데이터 분석 (0)	2021.11.03
16. EDA, 멕시코식당 주문 CHIPOTLE (0)	2021.10.28
15. 스크래핑 (0)	2021.10.28
14. Stockprice (2 (0)	2021.10.26

PREV 1 ···25 26 27 28 29 30 31 ···43 NEXT

전체 글

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

Image-to-Image Translation with Conditional Adversarial Networks

Conditional GAN

Generator

Discriminator

PatchGAN discriminator

cGAN의 목표

Pix2Pix 구현

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

Transfer learning 정리

U-net 구현하기

Concatenate vs Add

GAN(Generative Adversarial Network)

GAN 구조

GAN의 종류

Train_on_batch

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

Segmentation

FCN(Fully Convolutional Networks)

Skip combining

U-net

U-net 구조

Overlap-Tile Input

U-net 구현하기

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

Variational AutoEncoder

Reparametization trick

Regularizer term

Lasso

Ridge

Regularizer / penalty

loss function의 차이

'Computer_Science > Visual Intelligence' 카테고리의 다른 글

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

'Data_Science > Data_Analysis_Py' 카테고리의 다른 글

티스토리툴바