我正试图弄清楚如何保持删除某些层,以使模型达到block3_pool
(MaxPooling2D(。
IMAGE_SIZE = [32, 32]
model = VGG16(input_shape=IMAGE_SIZE + [3], weights='imagenet', include_top=False, classes=10)
我尝试了model.layers = model.layers[:-3]
,但它不起作用,而且model._layers.pop()
也不影响模型输出。
layer_dict = dict([(layer.name, layer) for layer in model_vgg.layers])
outputa = layer_dict['block3_pool'].output
new_model = tf.keras.Model(inputs=model_vgg.input, outputs=outputa)