ValueError：layersequential_1 的输入 0 与 layer 不兼容：：预期 min_ndim=4，发现 ndim=3。 收到的完整形状：[无、256、256]

Question

在我将图像转换为灰度之前，一切都很好。 所以 rgb 的形状是(256, 256, 3)但灰度有(256, 256) 。 当我喂它时，我得到了那个错误。

network = Sequential()

network.add(Convolution2D(32, kernel_size=(3, 3),strides=1,activation='relu',input_shape=(256, 256)))
network.add(MaxPooling2D((2, 2)))

# network.add(Convolution2D(32, kernel_size=(3, 3), strides=1, activation='relu'))
# network.add(MaxPooling2D((2, 2)))


network.add(Convolution2D(64, kernel_size=(3, 3), strides=1, activation='relu'))
network.add(MaxPooling2D((2, 2)))

# network.add(Convolution2D(64, kernel_size=(3, 3), strides=1, activation='relu'))
# network.add(MaxPooling2D((2, 2)))


network.add(Convolution2D(128, kernel_size=(3, 3), strides=1, activation='relu'))
network.add(MaxPooling2D((2, 2)))

# network.add(Convolution2D(128, kernel_size=(3, 3), strides=1, activation='relu'))
# network.add(MaxPooling2D((2, 2)))


network.add(Flatten())
network.add(Dense(256, activation = 'relu'))
network.add(Dense(2, activation = 'softmax'))

checkpoint_path = os.path.join("/---------/grayscale", "weights.best.hdf5")
checkpoint = ModelCheckpoint(checkpoint_path, monitor='val_accuracy', verbose=1, save_best_only=True, mode='max')
es = EarlyStopping(monitor='val_loss', mode='min', verbose=1, patience=10)
callbacks_list = [checkpoint, es]

network.compile(optimizer = 'adam', loss = 'categorical_crossentropy', metrics = ['accuracy'])

Answer 1

您必须在网络中提供形状为 256x256x1 的图像。

要将您的初始x_train转换为您的新X_train ：

X_train=np.reshape(x_train,(x_train.shape[0], x_train.shape[1],x_train.shape[2],1))

最后将您的 input_shape 从input_shape=(256,256)更改为input_shape=(256,256,1)