池化层和前卷积层的深度应相同。但不一样，请告诉我解决办法

Question

model = Sequential()

model.add(keras.layers.InputLayer(input_shape=input_shape))

model.add(keras.layers.convolutional.Conv2D(filters, filtersize, strides=(1, 1), padding='valid', data_format="channels_last", activation='relu'))

model.summary()

输出摘要为：

_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
input_10 (InputLayer)        (None, 300, 300, 3)       0         
_________________________________________________________________
conv2d_16 (Conv2D)           (None, 296, 296, 10)      760       
_________________________________________________________________
max_pooling2d_13 (MaxPooling (None, 296, 148, 5)       0         
_________________________________________________________________

上面的conv2dØ16第10层是深度，作为Maxpooling第5层，怎么可能呢

Answer 1

您很可能正在池层中使用设置data_format='channels_first'

我看到您将'channels_last'添加到卷积层，但您可能没有将其添加到池层

如果要更改keras的默认设置，请找到<user>/.keras/keras.json文件并在那里进行更改