케라스 CONVNET

import os, shutil

original_dataset_dir = 'C:/train'

# 소규모 데이터셋을 저장할 디렉터리
base_dir = 'C:/cats_and_dogs_small'
if os.path.exists(base_dir):  # 반복적인 실행을 위해 디렉토리를 삭제합니다.
    shutil.rmtree(base_dir)   
os.mkdir(base_dir)

# 훈련, 검증, 테스트 분할을 위한 디렉터리
train_dir = os.path.join(base_dir, 'train')
os.mkdir(train_dir)
validation_dir = os.path.join(base_dir, 'validation')
os.mkdir(validation_dir)
test_dir = os.path.join(base_dir, 'test')
os.mkdir(test_dir)

# 훈련용 고양이 사진 디렉터리
train_cats_dir = os.path.join(train_dir, 'cats')
os.mkdir(train_cats_dir)

# 훈련용 강아지 사진 디렉터리
train_dogs_dir = os.path.join(train_dir, 'dogs')
os.mkdir(train_dogs_dir)

# 검증용 고양이 사진 디렉터리
validation_cats_dir = os.path.join(validation_dir, 'cats')
os.mkdir(validation_cats_dir)

# 검증용 강아지 사진 디렉터리
validation_dogs_dir = os.path.join(validation_dir, 'dogs')
os.mkdir(validation_dogs_dir)

# 테스트용 고양이 사진 디렉터리
test_cats_dir = os.path.join(test_dir, 'cats')
os.mkdir(test_cats_dir)

# 테스트용 강아지 사진 디렉터리
test_dogs_dir = os.path.join(test_dir, 'dogs')
os.mkdir(test_dogs_dir)

# 처음 1,000개의 고양이 이미지를 train_cats_dir에 복사합니다
fnames = ['cat.{}.jpg'.format(i) for i in range(1000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(train_cats_dir, fname)
    shutil.copyfile(src, dst)

# 다음 500개 고양이 이미지를 validation_cats_dir에 복사합니다
fnames = ['cat.{}.jpg'.format(i) for i in range(1000, 1500)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(validation_cats_dir, fname)
    shutil.copyfile(src, dst)

# 다음 500개 고양이 이미지를 test_cats_dir에 복사합니다
fnames = ['cat.{}.jpg'.format(i) for i in range(1500, 2000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(test_cats_dir, fname)
    shutil.copyfile(src, dst)

# 처음 1,000개의 강아지 이미지를 train_dogs_dir에 복사합니다
fnames = ['dog.{}.jpg'.format(i) for i in range(1000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(train_dogs_dir, fname)
    shutil.copyfile(src, dst)

# 다음 500개 강아지 이미지를 validation_dogs_dir에 복사합니다
fnames = ['dog.{}.jpg'.format(i) for i in range(1000, 1500)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(validation_dogs_dir, fname)
    shutil.copyfile(src, dst)

# 다음 500개 강아지 이미지를 test_dogs_dir에 복사합니다
fnames = ['dog.{}.jpg'.format(i) for i in range(1500, 2000)]
for fname in fnames:
    src = os.path.join(original_dataset_dir, fname)
    dst = os.path.join(test_dogs_dir, fname)
    shutil.copyfile(src, dst)

네트워크 구성

from keras import layers, models

model=models.Sequential()
model.add(layers.Conv2D(32, (3,3), activation = 'relu', input_shape=(150,150,3)))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(64, (3,3), activation = 'relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(128, (3,3), activation = 'relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(128, (3,3), activation = 'relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Flatten())
model.add(layers.Dense(512, activation='relu'))
model.add(layers.Dense(1, activation = 'sigmoid'))

model.summary()

Model: "sequential_3"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
conv2d_8 (Conv2D)            (None, 148, 148, 32)      896       
_________________________________________________________________
max_pooling2d_8 (MaxPooling2 (None, 74, 74, 32)        0         
_________________________________________________________________
conv2d_9 (Conv2D)            (None, 72, 72, 64)        18496     
_________________________________________________________________
max_pooling2d_9 (MaxPooling2 (None, 36, 36, 64)        0         
_________________________________________________________________
conv2d_10 (Conv2D)           (None, 34, 34, 128)       73856     
_________________________________________________________________
max_pooling2d_10 (MaxPooling (None, 17, 17, 128)       0         
_________________________________________________________________
conv2d_11 (Conv2D)           (None, 15, 15, 128)       147584    
_________________________________________________________________
max_pooling2d_11 (MaxPooling (None, 7, 7, 128)         0         
_________________________________________________________________
flatten_3 (Flatten)          (None, 6272)              0         
_________________________________________________________________
dense_5 (Dense)              (None, 512)               3211776   
_________________________________________________________________
dense_6 (Dense)              (None, 1)                 513       
=================================================================
Total params: 3,453,121
Trainable params: 3,453,121
Non-trainable params: 0
_________________________________________________________________

모델 훈련 설정

from keras import optimizers

model.compile(loss='binary_crossentropy',
             optimizer=optimizers.RMSprop(lr=1e-4),
             metrics=['acc'])

데이터 전처리

from keras.preprocessing.image import ImageDataGenerator

train_datagen = ImageDataGenerator(rescale=1./255)
test_datagen = ImageDataGenerator(rescale=1./255)

#이미지파일을 전처리된 배치텐서로 바꿔주는 제네레이터
train_generator = train_datagen.flow_from_directory(
                train_dir,
                target_size=(150,150),
                batch_size=20,
                class_mode='binary')

validation_generator = test_datagen.flow_from_directory(
                validation_dir,
                target_size=(150,150),
                batch_size=20,
                class_mode='binary')

Found 2000 images belonging to 2 classes.
Found 1000 images belonging to 2 classes.

for data_batch, labels_batch in train_generator:
    print('배치 데이터 크기: ', data_batch.shape)
    print('배치 레이블 크기: ', labels_batch.shape)
    break

배치 데이터 크기:  (20, 150, 150, 3)
배치 레이블 크기:  (20,)

모델 훈련

history = model.fit_generator(
    train_generator,
    steps_per_epoch=100,#train데이터가 2000개.
    epochs=30,
    validation_data = validation_generator,
    validation_steps=50) #위에서 검증제네레이터의 배치를 20개씩 했으므로 50번 반복해야 전체(1000)사용

Epoch 1/30
100/100 [==============================] - 54s 542ms/step - loss: 0.6493 - acc: 0.6235 - val_loss: 0.6758 - val_acc: 0.6100
Epoch 2/30
100/100 [==============================] - 59s 587ms/step - loss: 0.5981 - acc: 0.6860 - val_loss: 0.7680 - val_acc: 0.6440
Epoch 3/30
100/100 [==============================] - 60s 604ms/step - loss: 0.5637 - acc: 0.7095 - val_loss: 0.6008 - val_acc: 0.6760
Epoch 4/30
100/100 [==============================] - 71s 710ms/step - loss: 0.5370 - acc: 0.7245 - val_loss: 0.8597 - val_acc: 0.5580
Epoch 5/30
100/100 [==============================] - 66s 656ms/step - loss: 0.5105 - acc: 0.7440 - val_loss: 0.7412 - val_acc: 0.6850
Epoch 6/30
100/100 [==============================] - 57s 569ms/step - loss: 0.4803 - acc: 0.7750 - val_loss: 0.4612 - val_acc: 0.6990
Epoch 7/30
100/100 [==============================] - 57s 570ms/step - loss: 0.4633 - acc: 0.7765 - val_loss: 0.6596 - val_acc: 0.7060
Epoch 8/30
100/100 [==============================] - 57s 574ms/step - loss: 0.4347 - acc: 0.7955 - val_loss: 0.3556 - val_acc: 0.6900
Epoch 9/30
100/100 [==============================] - 58s 577ms/step - loss: 0.4131 - acc: 0.8150 - val_loss: 0.6390 - val_acc: 0.7260
Epoch 10/30
100/100 [==============================] - 57s 571ms/step - loss: 0.3899 - acc: 0.8265 - val_loss: 0.3724 - val_acc: 0.7170
Epoch 11/30
100/100 [==============================] - 57s 571ms/step - loss: 0.3711 - acc: 0.8310 - val_loss: 0.5413 - val_acc: 0.7270
Epoch 12/30
100/100 [==============================] - 57s 571ms/step - loss: 0.3530 - acc: 0.8465 - val_loss: 0.6112 - val_acc: 0.7320
Epoch 13/30
100/100 [==============================] - 57s 570ms/step - loss: 0.3255 - acc: 0.8610 - val_loss: 0.3573 - val_acc: 0.6850
Epoch 14/30
100/100 [==============================] - 58s 577ms/step - loss: 0.3023 - acc: 0.8795 - val_loss: 0.5398 - val_acc: 0.7220
Epoch 15/30
100/100 [==============================] - 57s 569ms/step - loss: 0.2851 - acc: 0.8855 - val_loss: 0.3826 - val_acc: 0.7430
Epoch 16/30
100/100 [==============================] - 57s 574ms/step - loss: 0.2550 - acc: 0.8990 - val_loss: 0.4806 - val_acc: 0.7360
Epoch 17/30
100/100 [==============================] - 57s 568ms/step - loss: 0.2451 - acc: 0.9015 - val_loss: 0.8120 - val_acc: 0.7270
Epoch 18/30
100/100 [==============================] - 57s 570ms/step - loss: 0.2185 - acc: 0.9200 - val_loss: 0.8893 - val_acc: 0.7420
Epoch 19/30
100/100 [==============================] - 58s 581ms/step - loss: 0.1961 - acc: 0.9275 - val_loss: 0.5893 - val_acc: 0.7460
Epoch 20/30
100/100 [==============================] - 57s 571ms/step - loss: 0.1765 - acc: 0.9385 - val_loss: 0.3419 - val_acc: 0.7400
Epoch 21/30
100/100 [==============================] - 57s 570ms/step - loss: 0.1609 - acc: 0.9440 - val_loss: 1.1847 - val_acc: 0.7440
Epoch 22/30
100/100 [==============================] - 57s 568ms/step - loss: 0.1411 - acc: 0.9560 - val_loss: 0.5061 - val_acc: 0.7500
Epoch 23/30
100/100 [==============================] - 57s 571ms/step - loss: 0.1259 - acc: 0.9580 - val_loss: 0.7856 - val_acc: 0.7430
Epoch 24/30
100/100 [==============================] - 57s 574ms/step - loss: 0.1075 - acc: 0.9670 - val_loss: 1.5615 - val_acc: 0.7460
Epoch 25/30
100/100 [==============================] - 57s 569ms/step - loss: 0.0930 - acc: 0.9745 - val_loss: 1.0389 - val_acc: 0.7530
Epoch 26/30
100/100 [==============================] - 57s 569ms/step - loss: 0.0897 - acc: 0.9725 - val_loss: 0.2108 - val_acc: 0.7270
Epoch 27/30
100/100 [==============================] - 57s 570ms/step - loss: 0.0732 - acc: 0.9765 - val_loss: 0.4359 - val_acc: 0.7350
Epoch 28/30
100/100 [==============================] - 57s 569ms/step - loss: 0.0589 - acc: 0.9860 - val_loss: 0.4736 - val_acc: 0.7590
Epoch 29/30
100/100 [==============================] - 57s 569ms/step - loss: 0.0513 - acc: 0.9860 - val_loss: 0.6719 - val_acc: 0.7570
Epoch 30/30
100/100 [==============================] - 57s 565ms/step - loss: 0.0414 - acc: 0.9905 - val_loss: 0.8831 - val_acc: 0.7390

model.save('cats_and_dogs_small_1.h5')

import matplotlib.pyplot as plt

acc = history.history['acc']
val_acc = history.history['val_acc']

loss = history.history['loss']
val_loss = history.history['val_loss']

epochs = range(1, len(acc)+1)

plt.plot(epochs , acc, 'bo', label='Training acc')#x, y, 줄, label
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.legend()

plt.figure()

plt.plot(epochs, loss, 'bo', label='Training loss')
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()

plt.show()

epoch = 5 정도부터 정확도 70정도로 변함이없음

#

데이터 증식

여러 랜덤한 변환으로 샘플의 갯수를 늘림

datagen= ImageDataGenerator(
    rotation_range=20,     #사진 회전 각도
    width_shift_range=0.1, #수평 평행 이동
    height_shift_range=0.1,#수직 평행 이동
    shear_range=0.1,       #전단변환 각도
    zoom_range=0.1,        #확대 범위
    horizontal_flip=True,  #이미지를 수평으로 뒤집음
    fill_mode = 'nearest') #회전이나 이동으로 인해 새롭게 생성해야 할 픽셀을 채울 전략

from keras.preprocessing import image #이미지 전처리 유틸 모듈

fnames = sorted([os.path.join(train_cats_dir, fname) for
                fname in os.listdir(train_cats_dir)])

img_path = fnames[892]

img = image.load_img(img_path, target_size=(150, 150)) #이미지 읽고 크기 변경

x= image.img_to_array(img)  # 150,150,3 크기의 넘파이 배열로 변환
x= x.reshape((1,) +x.shape) # 1,150,150,3 크기로 변환


i=0
#랜덤하게 변환된 이미지 배치 무한히 생성
for batch in datagen.flow(x, batch_size=1):
    plt.figure(i)
    imgplot = plt.imshow(image.array_to_img(batch[0]))
    i += 1
    if i%4==0:
        break
plt.show()

데이터 증식을 했지만 원본 이미지수가 적기 때문에 데이터간 상호연관이 큼

과대적합 억제를 위한 Dropout층 추가

model = models.Sequential()
model.add(layers.Conv2D(32, (3,3), activation='relu',
                       input_shape=(150,150,3)))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(64, (3,3), activation='relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(128, (3,3), activation='relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Conv2D(128, (3,3), activation='relu'))
model.add(layers.MaxPooling2D(2,2))
model.add(layers.Flatten())
model.add(layers.Dropout(0.5))
model.add(layers.Dense(512, activation='relu'))
model.add(layers.Dense(1, activation='sigmoid'))

model.compile(loss = 'binary_crossentropy',
             optimizer = optimizers.RMSprop(lr=1e-4),
             metrics=['acc'])

train_datagen = ImageDataGenerator(
    rescale=1./255,
    rotation_range=40,
    width_shift_range=0.2,
    height_shift_range=0.2,
    shear_range=0.2,
    zoom_range=0.2,
    horizontal_flip = True,)

test_datagen=ImageDataGenerator(rescale=1./255)

train_generator = train_datagen.flow_from_directory(
    train_dir,
    target_size=(150,150),
    batch_size=32,
    class_mode='binary')

validation_generator = test_datagen.flow_from_directory(
    validation_dir,
    target_size=(150,150),
    batch_size=32,
    class_mode='binary')

history=model.fit_generator(
    train_generator,
    steps_per_epoch=100,
    epochs=100,
    validation_data=validation_generator,
    validation_steps=50)

Found 2000 images belonging to 2 classes.
Found 1000 images belonging to 2 classes.
Epoch 1/100
100/100 [==============================] - 92s 925ms/step - loss: 0.6887 - acc: 0.5283 - val_loss: 0.7003 - val_acc: 0.5596
Epoch 2/100
100/100 [==============================] - 92s 925ms/step - loss: 0.6626 - acc: 0.5830 - val_loss: 0.6613 - val_acc: 0.5760
Epoch 3/100
100/100 [==============================] - 93s 925ms/step - loss: 0.6517 - acc: 0.5997 - val_loss: 0.6276 - val_acc: 0.6301
Epoch 4/100
100/100 [==============================] - 94s 940ms/step - loss: 0.6286 - acc: 0.6451 - val_loss: 0.6294 - val_acc: 0.6289
Epoch 5/100
100/100 [==============================] - 92s 922ms/step - loss: 0.6138 - acc: 0.6585 - val_loss: 0.5954 - val_acc: 0.6567
Epoch 6/100
100/100 [==============================] - 93s 932ms/step - loss: 0.6028 - acc: 0.6660 - val_loss: 0.4447 - val_acc: 0.6701
Epoch 7/100
100/100 [==============================] - 93s 931ms/step - loss: 0.5955 - acc: 0.6812 - val_loss: 0.6383 - val_acc: 0.6859
Epoch 8/100
100/100 [==============================] - 92s 918ms/step - loss: 0.5922 - acc: 0.6771 - val_loss: 0.7146 - val_acc: 0.6772
Epoch 9/100
100/100 [==============================] - 92s 917ms/step - loss: 0.5721 - acc: 0.6932 - val_loss: 0.7393 - val_acc: 0.6134
Epoch 10/100
100/100 [==============================] - 92s 916ms/step - loss: 0.5715 - acc: 0.7063 - val_loss: 0.5376 - val_acc: 0.6948
Epoch 11/100
100/100 [==============================] - 93s 928ms/step - loss: 0.5663 - acc: 0.7030 - val_loss: 0.6372 - val_acc: 0.7249
Epoch 12/100
100/100 [==============================] - 92s 922ms/step - loss: 0.5654 - acc: 0.6932 - val_loss: 0.5627 - val_acc: 0.6459
Epoch 13/100
100/100 [==============================] - 91s 913ms/step - loss: 0.5576 - acc: 0.7105 - val_loss: 0.4204 - val_acc: 0.7178
Epoch 14/100
100/100 [==============================] - 113s 1s/step - loss: 0.5539 - acc: 0.7162 - val_loss: 0.4409 - val_acc: 0.7487
Epoch 15/100
100/100 [==============================] - 137s 1s/step - loss: 0.5513 - acc: 0.7114 - val_loss: 0.5906 - val_acc: 0.7049
Epoch 16/100
100/100 [==============================] - 139s 1s/step - loss: 0.5382 - acc: 0.7279 - val_loss: 0.4660 - val_acc: 0.7481
Epoch 17/100
100/100 [==============================] - 137s 1s/step - loss: 0.5451 - acc: 0.7220 - val_loss: 0.4494 - val_acc: 0.7341
Epoch 18/100
100/100 [==============================] - 141s 1s/step - loss: 0.5201 - acc: 0.7424 - val_loss: 0.5684 - val_acc: 0.7500
Epoch 19/100
100/100 [==============================] - 139s 1s/step - loss: 0.5193 - acc: 0.7409 - val_loss: 0.4637 - val_acc: 0.7627
Epoch 20/100
100/100 [==============================] - 137s 1s/step - loss: 0.5231 - acc: 0.7364 - val_loss: 0.3562 - val_acc: 0.7332
Epoch 21/100
100/100 [==============================] - 31673s 317s/step - loss: 0.5229 - acc: 0.7406 - val_loss: 0.4830 - val_acc: 0.7678
Epoch 22/100
100/100 [==============================] - 111s 1s/step - loss: 0.5097 - acc: 0.7427 - val_loss: 0.5673 - val_acc: 0.7732
Epoch 23/100
100/100 [==============================] - 117s 1s/step - loss: 0.5033 - acc: 0.7657 - val_loss: 0.4790 - val_acc: 0.7354
Epoch 24/100
100/100 [==============================] - 113s 1s/step - loss: 0.4987 - acc: 0.7624 - val_loss: 0.4347 - val_acc: 0.7629
Epoch 25/100
100/100 [==============================] - 109s 1s/step - loss: 0.5046 - acc: 0.7503 - val_loss: 0.4750 - val_acc: 0.7384
Epoch 26/100
100/100 [==============================] - 109s 1s/step - loss: 0.5043 - acc: 0.7547 - val_loss: 0.5126 - val_acc: 0.7538
Epoch 27/100
100/100 [==============================] - 109s 1s/step - loss: 0.4973 - acc: 0.7622 - val_loss: 0.3817 - val_acc: 0.7713
Epoch 28/100
100/100 [==============================] - 108s 1s/step - loss: 0.4907 - acc: 0.7652 - val_loss: 0.4684 - val_acc: 0.7786
Epoch 29/100
100/100 [==============================] - 108s 1s/step - loss: 0.4908 - acc: 0.7667 - val_loss: 0.3464 - val_acc: 0.7281
Epoch 30/100
100/100 [==============================] - 111s 1s/step - loss: 0.4846 - acc: 0.7660 - val_loss: 0.3983 - val_acc: 0.7582
Epoch 31/100
100/100 [==============================] - 108s 1s/step - loss: 0.4824 - acc: 0.7654 - val_loss: 0.3176 - val_acc: 0.7841
Epoch 32/100
100/100 [==============================] - 106s 1s/step - loss: 0.4890 - acc: 0.7614 - val_loss: 0.4906 - val_acc: 0.7094
Epoch 33/100
100/100 [==============================] - 110s 1s/step - loss: 0.4720 - acc: 0.7660 - val_loss: 0.5354 - val_acc: 0.7931
Epoch 34/100
100/100 [==============================] - 107s 1s/step - loss: 0.4662 - acc: 0.7721 - val_loss: 0.3767 - val_acc: 0.7957
Epoch 35/100
100/100 [==============================] - 112s 1s/step - loss: 0.4815 - acc: 0.7661 - val_loss: 0.5181 - val_acc: 0.8001
Epoch 36/100
100/100 [==============================] - 127s 1s/step - loss: 0.4580 - acc: 0.7871 - val_loss: 0.3505 - val_acc: 0.7893
Epoch 37/100
100/100 [==============================] - 114s 1s/step - loss: 0.4707 - acc: 0.7648 - val_loss: 0.3029 - val_acc: 0.7868
Epoch 38/100
100/100 [==============================] - 137s 1s/step - loss: 0.4645 - acc: 0.7833 - val_loss: 0.4289 - val_acc: 0.7706
Epoch 39/100
100/100 [==============================] - 138s 1s/step - loss: 0.4624 - acc: 0.7839 - val_loss: 0.6486 - val_acc: 0.7595
Epoch 40/100
100/100 [==============================] - 124s 1s/step - loss: 0.4413 - acc: 0.7905 - val_loss: 0.5328 - val_acc: 0.7719
Epoch 41/100
100/100 [==============================] - 127s 1s/step - loss: 0.4578 - acc: 0.7863 - val_loss: 0.7052 - val_acc: 0.7700
Epoch 42/100
100/100 [==============================] - 130s 1s/step - loss: 0.4422 - acc: 0.7926 - val_loss: 0.5355 - val_acc: 0.7957
Epoch 43/100
100/100 [==============================] - 127s 1s/step - loss: 0.4352 - acc: 0.8018 - val_loss: 0.5226 - val_acc: 0.7764
Epoch 44/100
100/100 [==============================] - 117s 1s/step - loss: 0.4456 - acc: 0.7937 - val_loss: 0.5422 - val_acc: 0.7862
Epoch 45/100
100/100 [==============================] - 140s 1s/step - loss: 0.4314 - acc: 0.8002 - val_loss: 0.2976 - val_acc: 0.7906
Epoch 46/100
100/100 [==============================] - 138s 1s/step - loss: 0.4355 - acc: 0.7964 - val_loss: 0.3651 - val_acc: 0.7951
Epoch 47/100
100/100 [==============================] - 129s 1s/step - loss: 0.4251 - acc: 0.8033 - val_loss: 0.3866 - val_acc: 0.7545
Epoch 48/100
100/100 [==============================] - 122s 1s/step - loss: 0.4355 - acc: 0.7993 - val_loss: 0.3938 - val_acc: 0.7957
Epoch 49/100
100/100 [==============================] - 126s 1s/step - loss: 0.4212 - acc: 0.8103 - val_loss: 0.4630 - val_acc: 0.7836
Epoch 50/100
100/100 [==============================] - 136s 1s/step - loss: 0.4326 - acc: 0.7992 - val_loss: 0.2479 - val_acc: 0.8177
Epoch 51/100
100/100 [==============================] - 131s 1s/step - loss: 0.4219 - acc: 0.8043 - val_loss: 0.4381 - val_acc: 0.7874
Epoch 52/100
100/100 [==============================] - 134s 1s/step - loss: 0.4237 - acc: 0.8052 - val_loss: 0.3356 - val_acc: 0.7500
Epoch 53/100
100/100 [==============================] - 128s 1s/step - loss: 0.4218 - acc: 0.8068 - val_loss: 0.6825 - val_acc: 0.7735
Epoch 54/100
100/100 [==============================] - 132s 1s/step - loss: 0.4140 - acc: 0.8084 - val_loss: 0.7228 - val_acc: 0.8119
Epoch 55/100
100/100 [==============================] - 133s 1s/step - loss: 0.4238 - acc: 0.8009 - val_loss: 0.4984 - val_acc: 0.8160
Epoch 56/100
100/100 [==============================] - 119s 1s/step - loss: 0.4142 - acc: 0.8144 - val_loss: 0.6107 - val_acc: 0.7996
Epoch 57/100
100/100 [==============================] - 116s 1s/step - loss: 0.4121 - acc: 0.8103 - val_loss: 0.4271 - val_acc: 0.8015
Epoch 58/100
100/100 [==============================] - 115s 1s/step - loss: 0.4109 - acc: 0.8122 - val_loss: 0.5391 - val_acc: 0.7995
Epoch 59/100
100/100 [==============================] - 94s 940ms/step - loss: 0.3982 - acc: 0.8153 - val_loss: 0.2988 - val_acc: 0.8170
Epoch 60/100
100/100 [==============================] - 99s 987ms/step - loss: 0.4110 - acc: 0.8125 - val_loss: 0.3002 - val_acc: 0.7824
Epoch 61/100
100/100 [==============================] - 92s 923ms/step - loss: 0.4004 - acc: 0.8200 - val_loss: 0.4444 - val_acc: 0.8215
Epoch 62/100
100/100 [==============================] - 93s 929ms/step - loss: 0.4038 - acc: 0.8150 - val_loss: 0.5618 - val_acc: 0.7874
Epoch 63/100
100/100 [==============================] - 92s 923ms/step - loss: 0.3998 - acc: 0.8213 - val_loss: 0.3710 - val_acc: 0.7899
Epoch 64/100
100/100 [==============================] - 91s 911ms/step - loss: 0.3910 - acc: 0.8172 - val_loss: 0.7399 - val_acc: 0.7996
Epoch 65/100
100/100 [==============================] - 92s 923ms/step - loss: 0.3966 - acc: 0.8207 - val_loss: 0.5853 - val_acc: 0.8115
Epoch 66/100
100/100 [==============================] - 92s 918ms/step - loss: 0.3904 - acc: 0.8191 - val_loss: 0.5106 - val_acc: 0.7784
Epoch 67/100
100/100 [==============================] - 92s 923ms/step - loss: 0.3853 - acc: 0.8239 - val_loss: 0.5523 - val_acc: 0.8071
Epoch 68/100
100/100 [==============================] - 92s 919ms/step - loss: 0.3860 - acc: 0.8286 - val_loss: 0.6057 - val_acc: 0.8177
Epoch 69/100
100/100 [==============================] - 94s 936ms/step - loss: 0.3923 - acc: 0.8239 - val_loss: 0.3772 - val_acc: 0.8147
Epoch 70/100
100/100 [==============================] - 92s 924ms/step - loss: 0.3770 - acc: 0.8320 - val_loss: 0.5353 - val_acc: 0.8222
Epoch 71/100
100/100 [==============================] - 91s 913ms/step - loss: 0.3784 - acc: 0.8318 - val_loss: 0.4160 - val_acc: 0.8261
Epoch 72/100
100/100 [==============================] - 92s 925ms/step - loss: 0.3635 - acc: 0.8351 - val_loss: 0.5472 - val_acc: 0.8235
Epoch 73/100
100/100 [==============================] - 91s 910ms/step - loss: 0.3740 - acc: 0.8299 - val_loss: 0.6288 - val_acc: 0.7738
Epoch 74/100
100/100 [==============================] - 93s 926ms/step - loss: 0.3785 - acc: 0.8311 - val_loss: 0.2810 - val_acc: 0.8147
Epoch 75/100
100/100 [==============================] - 91s 914ms/step - loss: 0.3648 - acc: 0.8339 - val_loss: 0.2999 - val_acc: 0.8293
Epoch 76/100
100/100 [==============================] - 91s 905ms/step - loss: 0.3730 - acc: 0.8349 - val_loss: 0.4934 - val_acc: 0.7881
Epoch 77/100
100/100 [==============================] - 92s 923ms/step - loss: 0.3633 - acc: 0.8455 - val_loss: 0.4755 - val_acc: 0.8086
Epoch 78/100
100/100 [==============================] - 91s 908ms/step - loss: 0.3749 - acc: 0.8302 - val_loss: 0.5254 - val_acc: 0.7443
Epoch 79/100
100/100 [==============================] - 92s 922ms/step - loss: 0.3677 - acc: 0.8361 - val_loss: 0.6599 - val_acc: 0.7661
Epoch 80/100
100/100 [==============================] - 90s 903ms/step - loss: 0.3610 - acc: 0.8409 - val_loss: 0.4462 - val_acc: 0.8119
Epoch 81/100
100/100 [==============================] - 92s 925ms/step - loss: 0.3615 - acc: 0.8384 - val_loss: 0.4108 - val_acc: 0.8084
Epoch 82/100
100/100 [==============================] - 92s 917ms/step - loss: 0.3466 - acc: 0.8489 - val_loss: 0.3377 - val_acc: 0.8028
Epoch 83/100
100/100 [==============================] - 90s 905ms/step - loss: 0.3570 - acc: 0.8415 - val_loss: 0.3121 - val_acc: 0.8261
Epoch 84/100
100/100 [==============================] - 92s 918ms/step - loss: 0.3508 - acc: 0.8464 - val_loss: 0.7422 - val_acc: 0.8048
Epoch 85/100
100/100 [==============================] - 91s 906ms/step - loss: 0.3523 - acc: 0.8456 - val_loss: 0.3180 - val_acc: 0.8223
Epoch 86/100
100/100 [==============================] - 92s 923ms/step - loss: 0.3625 - acc: 0.8359 - val_loss: 0.4887 - val_acc: 0.8099
Epoch 87/100
100/100 [==============================] - 3650s 37s/step - loss: 0.3411 - acc: 0.8492 - val_loss: 0.3528 - val_acc: 0.7633
Epoch 88/100
100/100 [==============================] - 104s 1s/step - loss: 0.3558 - acc: 0.8456 - val_loss: 0.2655 - val_acc: 0.8396
Epoch 89/100
100/100 [==============================] - 100s 1s/step - loss: 0.3324 - acc: 0.8502 - val_loss: 0.3610 - val_acc: 0.8177
Epoch 90/100
100/100 [==============================] - 103s 1s/step - loss: 0.3557 - acc: 0.8403 - val_loss: 0.3775 - val_acc: 0.8255
Epoch 91/100
100/100 [==============================] - 110s 1s/step - loss: 0.3381 - acc: 0.8494 - val_loss: 0.5460 - val_acc: 0.8370
Epoch 92/100
100/100 [==============================] - 107s 1s/step - loss: 0.3438 - acc: 0.8436 - val_loss: 0.4710 - val_acc: 0.8338
Epoch 93/100
100/100 [==============================] - 105s 1s/step - loss: 0.3300 - acc: 0.8549 - val_loss: 0.3508 - val_acc: 0.8370
Epoch 94/100
100/100 [==============================] - 107s 1s/step - loss: 0.3310 - acc: 0.8516 - val_loss: 0.2825 - val_acc: 0.8319
Epoch 95/100
100/100 [==============================] - 105s 1s/step - loss: 0.3412 - acc: 0.8510 - val_loss: 0.3124 - val_acc: 0.8376
Epoch 96/100
100/100 [==============================] - 107s 1s/step - loss: 0.3275 - acc: 0.8551 - val_loss: 0.1276 - val_acc: 0.8286
Epoch 97/100
100/100 [==============================] - 105s 1s/step - loss: 0.3362 - acc: 0.8552 - val_loss: 0.3917 - val_acc: 0.8331
Epoch 98/100
100/100 [==============================] - 107s 1s/step - loss: 0.3110 - acc: 0.8608 - val_loss: 0.4124 - val_acc: 0.8254
Epoch 99/100
100/100 [==============================] - 106s 1s/step - loss: 0.3344 - acc: 0.8496 - val_loss: 0.3453 - val_acc: 0.8115
Epoch 100/100
100/100 [==============================] - 104s 1s/step - loss: 0.3332 - acc: 0.8573 - val_loss: 0.3846 - val_acc: 0.8177

model.save('cats_and_dogs_small_2.h5')

acc=history.history['acc']
val_acc=history.history['val_acc']
loss=history.history['loss']
val_loss=history.history['val_loss']

epochs=range(1,len(acc)+1)

plt.plot(epochs, acc, 'bo', label='Training acc')
plt.plot(epochs, val_acc, 'b', label='Validation acc')
plt.title('Training and validation accuracy')
plt.legend()

plt.figure()#위에 이미 figure가 있으니 새로운 figure선언하여 새 그래프를 그릴수있게함

plt.plot(epochs, loss, 'bo', label='Training loss')
plt.plot(epochs, val_loss, 'b', label='Validation loss')
plt.title('Training and validation loss')
plt.legend()

<matplotlib.legend.Legend at 0x1de3976fc48>