首先，我是CNN和Keras的新手。我正在尝试创建基于U-Net架构的神经网络。这是使用Keras的python模型：
inputs = Input((512, 512, 11))
conv1 = Conv2D(64, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(inputs)
conv1 = Conv2D(64, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv1)
pool1 = MaxPooling2D(pool_size=(2, 2))(conv1)
pool1 = Dropout(0.1)(pool1)
conv2 = Conv2D(128, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool1)
conv2 = Conv2D(128, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv2)
pool2 = MaxPooling2D(pool_size=(2, 2))(conv2)
pool2 = Dropout(0.1)(pool2)
conv3 = Conv2D(256, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool2)
conv3 = Conv2D(256, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv3)
pool3 = MaxPooling2D(pool_size=(2, 2))(conv3)
pool3 = Dropout(0.1)(pool3)
conv4 = Conv2D(512, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool3)
conv4 = Conv2D(512, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv4)
pool4 = MaxPooling2D(pool_size=(2, 2))(conv4)
drop4 = Dropout(0.1)(conv4)
conv5 = Conv2D(1024, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(pool4)
conv5 = Conv2D(1024, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv5)
up6 = Conv2DTranspose(512, 2, strides = (2, 2), activation='relu', padding='same', kernel_initializer='he_normal')(conv5)
up6 = concatenate([drop4,up6])
up6 = Dropout(0.1)(up6)
conv6 = Conv2D(512, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(up6)
conv6 = Conv2D(512, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv6)
up7 = Conv2DTranspose(256, 2, strides = (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv6)
up7 = concatenate([conv3,up7])
up7 = Dropout(0.1)(up7)
conv7 = Conv2D(256, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(up7)
conv7 = Conv2D(256, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv7)
up8 = Conv2DTranspose(128, 2, strides = (2, 2), activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv7)
up8 = concatenate([conv2,up8])
up8 = Dropout(0.1)(up8)
conv8 = Conv2D(128, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(up8)
conv8 = Conv2D(128, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv8)
up9 = Conv2DTranspose(64, 2, strides = (2, 2),activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv8)
up9 = concatenate([conv1,up9])
up9 = Dropout(0.1)(up9)
conv9 = Conv2D(64, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(up9)
conv9 = Conv2D(64, 3, activation = 'relu', padding = 'same', kernel_initializer = 'he_normal')(conv9)
conv10 = Conv2D(1, 1, activation = 'sigmoid')(conv9)
model = Model(input = [inputs], output = [conv10])
model.compile(optimizer = Adam(lr = 1e-4), loss = 'binary_crossentropy', metrics = ['accuracy'])
return model
model.summary（）
Layer (type) Output Shape Param # Connected to
==================================================================================================
input_1 (InputLayer) (None, 512, 512, 11) 0
__________________________________________________________________________________________________
conv2d_1 (Conv2D) (None, 512, 512, 64) 6400 input_1[0][0]
__________________________________________________________________________________________________
conv2d_2 (Conv2D) (None, 512, 512, 64) 36928 conv2d_1[0][0]
__________________________________________________________________________________________________
max_pooling2d_1 (MaxPooling2D) (None, 256, 256, 64) 0 conv2d_2[0][0]
__________________________________________________________________________________________________
dropout_1 (Dropout) (None, 256, 256, 64) 0 max_pooling2d_1[0][0]
__________________________________________________________________________________________________
conv2d_3 (Conv2D) (None, 256, 256, 128 73856 dropout_1[0][0]
__________________________________________________________________________________________________
conv2d_4 (Conv2D) (None, 256, 256, 128 147584 conv2d_3[0][0]
__________________________________________________________________________________________________
max_pooling2d_2 (MaxPooling2D) (None, 128, 128, 128 0 conv2d_4[0][0]
__________________________________________________________________________________________________
dropout_2 (Dropout) (None, 128, 128, 128 0 max_pooling2d_2[0][0]
__________________________________________________________________________________________________
conv2d_5 (Conv2D) (None, 128, 128, 256 295168 dropout_2[0][0]
__________________________________________________________________________________________________
conv2d_6 (Conv2D) (None, 128, 128, 256 590080 conv2d_5[0][0]
__________________________________________________________________________________________________
max_pooling2d_3 (MaxPooling2D) (None, 64, 64, 256) 0 conv2d_6[0][0]
__________________________________________________________________________________________________
dropout_3 (Dropout) (None, 64, 64, 256) 0 max_pooling2d_3[0][0]
__________________________________________________________________________________________________
conv2d_7 (Conv2D) (None, 64, 64, 512) 1180160 dropout_3[0][0]
__________________________________________________________________________________________________
conv2d_8 (Conv2D) (None, 64, 64, 512) 2359808 conv2d_7[0][0]
__________________________________________________________________________________________________
max_pooling2d_4 (MaxPooling2D) (None, 32, 32, 512) 0 conv2d_8[0][0]
__________________________________________________________________________________________________
conv2d_9 (Conv2D) (None, 32, 32, 1024) 4719616 max_pooling2d_4[0][0]
__________________________________________________________________________________________________
conv2d_10 (Conv2D) (None, 32, 32, 1024) 9438208 conv2d_9[0][0]
__________________________________________________________________________________________________
dropout_4 (Dropout) (None, 64, 64, 512) 0 conv2d_8[0][0]
__________________________________________________________________________________________________
conv2d_transpose_1 (Conv2DTrans (None, 64, 64, 512) 2097664 conv2d_10[0][0]
__________________________________________________________________________________________________
concatenate_1 (Concatenate) (None, 64, 64, 1024) 0 dropout_4[0][0]
conv2d_transpose_1[0][0]
__________________________________________________________________________________________________
dropout_5 (Dropout) (None, 64, 64, 1024) 0 concatenate_1[0][0]
__________________________________________________________________________________________________
conv2d_11 (Conv2D) (None, 64, 64, 512) 4719104 dropout_5[0][0]
__________________________________________________________________________________________________
conv2d_12 (Conv2D) (None, 64, 64, 512) 2359808 conv2d_11[0][0]
__________________________________________________________________________________________________
conv2d_transpose_2 (Conv2DTrans (None, 128, 128, 256 524544 conv2d_12[0][0]
__________________________________________________________________________________________________
concatenate_2 (Concatenate) (None, 128, 128, 512 0 conv2d_6[0][0]
conv2d_transpose_2[0][0]
__________________________________________________________________________________________________
dropout_6 (Dropout) (None, 128, 128, 512 0 concatenate_2[0][0]
__________________________________________________________________________________________________
conv2d_13 (Conv2D) (None, 128, 128, 256 1179904 dropout_6[0][0]
__________________________________________________________________________________________________
conv2d_14 (Conv2D) (None, 128, 128, 256 590080 conv2d_13[0][0]
__________________________________________________________________________________________________
conv2d_transpose_3 (Conv2DTrans (None, 256, 256, 128 131200 conv2d_14[0][0]
__________________________________________________________________________________________________
concatenate_3 (Concatenate) (None, 256, 256, 256 0 conv2d_4[0][0]
conv2d_transpose_3[0][0]
__________________________________________________________________________________________________
dropout_7 (Dropout) (None, 256, 256, 256 0 concatenate_3[0][0]
__________________________________________________________________________________________________
conv2d_15 (Conv2D) (None, 256, 256, 128 295040 dropout_7[0][0]
__________________________________________________________________________________________________
conv2d_16 (Conv2D) (None, 256, 256, 128 147584 conv2d_15[0][0]
__________________________________________________________________________________________________
conv2d_transpose_4 (Conv2DTrans (None, 512, 512, 64) 32832 conv2d_16[0][0]
__________________________________________________________________________________________________
concatenate_4 (Concatenate) (None, 512, 512, 128 0 conv2d_2[0][0]
conv2d_transpose_4[0][0]
__________________________________________________________________________________________________
dropout_8 (Dropout) (None, 512, 512, 128 0 concatenate_4[0][0]
__________________________________________________________________________________________________
conv2d_17 (Conv2D) (None, 512, 512, 64) 73792 dropout_8[0][0]
__________________________________________________________________________________________________
conv2d_18 (Conv2D) (None, 512, 512, 64) 36928 conv2d_17[0][0]
__________________________________________________________________________________________________
conv2d_19 (Conv2D) (None, 512, 512, 1) 65 conv2d_18[0][0]
==================================================================================================
Total params: 31,036,353
Trainable params: 31,036,353
Non-trainable params: 0
要训练模型，我使用两个numpy数组：
x_train，每个元素的形状都为（512、512、11）
y_train，每个元素的形状都为（512，512）
我的问题是conv2d_19的形状输出为（512，512，1）
我正在寻找丢弃conv2d_19输出中最后一个维的正确方法，以解决错误，而不必更改y_train中每个元素的维。我不知道我的解释是否正确。
这是错误：
Traceback (most recent call last):
File "main.py", line 14, in <module>
history = model.fit(x_train, y_train, batch_size=32, epochs=1)
File "/home/luis/.local/lib/python3.6/site-packages/keras/engine/training.py", line 1154, in fit
batch_size=batch_size)
File "/home/luis/.local/lib/python3.6/site-packages/keras/engine/training.py", line 621, in _standardize_user_data
exception_prefix='target')
File "/home/luis/.local/lib/python3.6/site-packages/keras/engine/training_utils.py", line 135, in standardize_input_data
'with shape ' + str(data_shape))
ValueError: Error when checking target: expected conv2d_19 to have 4 dimensions, but got array with shape (32, 512, 512)
答案 0 :(得分：0)
使y_train与模型的输出具有相同的形状。那就是(512, 512, 1)。
您的y_train当前为(512,512)
y_train = y_train.reshape((512,512,1))
或者在模型中
....
conv10 = Conv2D(1, 1, activation = 'sigmoid')(conv9)
conv10 = Reshape((512,512))(conv10)
....
答案 1 :(得分：0)
来自https://keras.io/backend/#reshape
from keras import backend as K
out = K.reshape(conv10, (512,512))
然后使用out而不是conv10编译模型