더북(TheBook)

앞서 생성한 클래스(VGG19)를 호출하여 VGG19라는 모델을 생성합니다. 이때 VGG19 클래스에 전달되는 입력 값은 (224,224,3)의 형태를 갖습니다.

코드 6-16 VGG19 모델 출력

model = VGG19(input_shape=(224,224,3))
model.summary()

다음은 VGG19 모델의 출력 결과입니다.

Model: "vg_g19"
________________________________________________________________
Layer (type)                 Output Shape              Param #
================================================================
conv2d (Conv2D)              (None, 224, 224, 64)      1792
________________________________________________________________
conv2d_1 (Conv2D)            (None, 224, 224, 64)      36928
________________________________________________________________
max_pooling2d (MaxPooling2D) (None, 112, 112, 64)      0
________________________________________________________________
conv2d_2 (Conv2D)            (None, 112, 112, 128)     73856
________________________________________________________________
conv2d_3 (Conv2D)            (None, 112, 112, 128)     147584
________________________________________________________________
max_pooling2d_1 (MaxPooling2 (None, 56, 56, 128)       0
________________________________________________________________
conv2d_4 (Conv2D)            (None, 56, 56, 256)       295168
________________________________________________________________
conv2d_5 (Conv2D)            (None, 56, 56, 256)       590080
________________________________________________________________
conv2d_6 (Conv2D)            (None, 56, 56, 256)       590080
________________________________________________________________
conv2d_7 (Conv2D)            (None, 56, 56, 256)       590080
________________________________________________________________
max_pooling2d_2 (MaxPooling2 (None, 28, 28, 256)       0
________________________________________________________________
conv2d_8 (Conv2D)            (None, 28, 28, 512)       1180160
________________________________________________________________
conv2d_9 (Conv2D)            (None, 28, 28, 512)       2359808
________________________________________________________________
conv2d_10 (Conv2D)           (None, 28, 28, 512)       2359808
________________________________________________________________
conv2d_11 (Conv2D)           (None, 28, 28, 512)       2359808
________________________________________________________________
max_pooling2d_3 (MaxPooling2 (None, 14, 14, 512)       0
________________________________________________________________
conv2d_12 (Conv2D)           (None, 14, 14, 512)       2359808
________________________________________________________________
conv2d_13 (Conv2D)           (None, 14, 14, 512)       2359808
________________________________________________________________
conv2d_14 (Conv2D)           (None, 14, 14, 512)       2359808
________________________________________________________________
conv2d_15 (Conv2D)           (None, 14, 14, 512)       2359808
________________________________________________________________
max_pooling2d_4 (MaxPooling2 (None, 7, 7, 512)         0
________________________________________________________________
flatten (Flatten)            (None, 25088)             0
________________________________________________________________
dense (Dense)                (None, 4096)              102764544
________________________________________________________________
dropout (Dropout)            (None, 4096)              0
________________________________________________________________
dense_1 (Dense)              (None, 4096)              16781312
________________________________________________________________
dropout_1 (Dropout)          (None, 4096)              0
________________________________________________________________
dense_2 (Dense)              (None, 1000)              4097000
================================================================
Total params: 143,667,240
Trainable params: 143,667,240
Non-trainable params: 0
________________________________________________________________
신간 소식 구독하기
뉴스레터에 가입하시고 이메일로 신간 소식을 받아 보세요.