AttributeError: 'Concatenate' object has no attribute 'shape'

Question

我正在尝试在 tensorflow 中进行一些图像分割，这是我的模型：

inputs = Input((IMAGE_HEIGHT, IMAGE_WIDTH, 3))
s = Lambda(lambda x: x / 255) (inputs)

conv1 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (inputs)
conv1 = BatchNormalization() (conv1)
conv1 = Dropout(0.1) (conv1)
conv1 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv1)
conv1 = BatchNormalization() (conv1)
pooling1 = MaxPooling2D((2, 2)) (conv1)

conv2 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (pooling1)
conv2 = BatchNormalization() (conv2)
conv2 = Dropout(0.1) (conv2)
conv2 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv2)
conv2 = BatchNormalization() (conv2)
pooling2 = MaxPooling2D((2, 2)) (conv2)

conv3 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (pooling2)
conv3 = BatchNormalization() (conv3)
conv3 = Dropout(0.2) (conv3)
conv3 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv3)
conv3 = BatchNormalization() (conv3)
pooling3 = MaxPooling2D((2, 2)) (conv3)

conv4 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (pooling3)
conv4 = BatchNormalization() (conv4)
conv4 = Dropout(0.2) (conv4)
conv4 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv4)
conv4 = BatchNormalization() (conv4)
pooling4 = MaxPooling2D(pool_size=(2, 2)) (conv4)

conv5 = Conv2D(256, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (pooling4)
conv5 = BatchNormalization() (conv5)
conv5 = Dropout(0.3) (conv5)
conv5 = Conv2D(256, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv5)
conv5 = BatchNormalization() (conv5)


upsample6 = Conv2DTranspose(128, (2, 2), strides=(2, 2), padding='same') (conv5)
upsample6 = Concatenate([upsample6, conv4])
conv6 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (upsample6)
conv6 = BatchNormalization() (conv6)
conv6 = Dropout(0.2) (conv6)
conv6 = Conv2D(128, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv6)
conv6 = BatchNormalization() (conv6)

upsample7 = Conv2DTranspose(64, (2, 2), strides=(2, 2), padding='same') (conv6)
upsample7 = Concatenate([upsample7, conv3])
conv7 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (upsample7)
conv7 = BatchNormalization() (conv7)
conv7 = Dropout(0.2) (conv7)
conv7 = Conv2D(64, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv7)
conv7 = BatchNormalization() (conv7)

upsample8 = Conv2DTranspose(32, (2, 2), strides=(2, 2), padding='same') (conv7)
upsample8 = Concatenate([upsample8, conv2])
conv8 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (upsample8)
conv8 = BatchNormalization() (conv8)
conv8 = Dropout(0.1) (conv8)
conv8 = Conv2D(32, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv8)
conv8 = BatchNormalization() (conv8)

upsample9 = Conv2DTranspose(16, (2, 2), strides=(2, 2), padding='same') (conv8)
upsample9 = Concatenate([upsample9, conv1], axis=3)
conv9 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (upsample9)
conv9 = BatchNormalization() (conv9)
conv9 = Dropout(0.1) (conv9)
conv9 = Conv2D(16, (3, 3), activation='elu', kernel_initializer='he_normal', padding='same') (conv9)
conv9 = BatchNormalization() (conv9)

outputs = Conv2D(1, (1, 1), activation='sigmoid') (conv9)

model = Model(inputs=[inputs], outputs=[outputs])
model.summary()

但是它给我这个错误：

AttributeError: 'Concatenate' object has no attribute 'shape'

我的进口：

from tensorflow.keras.layers import Dense, Dropout, Lambda, Input, Masking,...
from tensorflow.keras.layers import Reshape, Dropout, Dense,Multiply, Dot, Concatenate,Embedding
...

我该如何解决这个问题？

来源 - https://github.com/Paulymorphous/skeyenet/blob/master/Src/Road_Detection_GPU.ipynb

由于我使用的是来自 tensorflow 的 keras，请以这种方式提出一些解决方案，否则它会破坏整个项目，我必须更改整个结构。

Answer 1

尝试使用：

upsample7 = Concatenate()([upsample7, conv3])

并更新每个上采样以及使用连接函数的位置。

注意 Concatenate 中的 () ^^