Tensorflow 如何使用Keras中的函数API在快捷连接中添加卷积层？

我正在尝试使用Keras中的函数API实现自定义CNN模型，如附图所示。我已经编写了实现主分支的代码，但是在添加1x1卷积作为快捷连接时遇到了问题。在最大池层之前，为每对卷积块添加快捷卷积。代码如下：

input_shape = (256,256,3)
model_input = Input(shape=input_shape)
print(model_input) 

def custom_cnn(model_input):
    x = Conv2D(16, (3, 3), strides = (2,2), padding = 'same')(model_input)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(16, (3, 3), strides = (2,2), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(32, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(32, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(48, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(48, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(64, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(64, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(80, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(80, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
            
    x = GlobalAveragePooling2D()(x)
    x = Dense(num_classes, activation='softmax')(x)
    model = Model(inputs=model_input, outputs=x, name='custom_cnn')
    return model

#instantiate the model
custom_model = custom_cnn(model_input)

#display model summary
custom_model.summary() 

---

您可以使用

keras.layers.concatenate

功能合并多个层，因此如果您有两个“并行”层

x

和

y

，您可以将它们合并到层

z

，如下所示：

z = layers.concatenate([x, y])

在您的代码中，它看起来如下：

z = layers.concatenate([x, y])

首先，创建一个“平行”层（注意，从

model_input

层开始，两个层

x

和以后的

y

将应用于同一层）

然后创建第二个“平行”层：

合并它们

z = layers.concatenate([x, y])

并在

z

上应用最大池（在连接的结果层上）

---

再次，创建两个分支，都将

z

作为输入（其中

z

相当于上一次迭代中的

model_输入

），并在需要应用串联的每个块上重复该操作。

这里，按照模式在网络中实现剩余块：

---

谢谢，但是在第一次连接之后它会抛出一个错误：ValueError:A

连接

层需要输入匹配的形状，concat轴除外。已获取输入形状：[（无，8，8，32），（无，31，31，32）]，因为不同输入维度的连接问题。作者的tensorflow代码位于，但我在将其转换为Keras兼容代码时遇到问题。这是因为要连接的层的维度必须相同，但执行连接的轴除外。您可以指定要连接的轴，但其他轴必须匹配，这与您的输出不同，其中两个维度不匹配（8！=31和8！=31）@shiva请告诉我是否有问题，谢谢，但此处显示的模型不只是有剩余连接，而是使用具有不同步长的1x1卷积。我在使用连接层时也遇到了问题，因为尺寸不匹配。作者在github.com/frapa/tbcnn/blob/master/net.py上有他们的tensorflow代码，但我在将它们转换为与Keras兼容的代码时遇到了问题。。。1x1是您可以看到的过滤器Conv2D（…，（1，1））。我提供的模型是图片中所示内容的精确复制品。连接错误。。。您必须操作剩余连接谢谢，但我无法理解剩余连接中使用的不同步幅值。例如，第一个剩余连接使用4个步幅，而其余的使用一个步幅。我使用图片中报告的相同精确值（第一个块：4，第二个块：1，3个块：1，4个块：1，5个块：1）。当你操作连接时，为了匹配维度，你必须这样做。第一个块连接输入，而其他块连接池，池由他们自己操作，减少（因此需要1步）

z = layers.concatenate([x, y])

z = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(z)

num_classes = 3
input_shape = (256,256,3)
model_input = Input(shape=input_shape)

def custom_cnn(model_input):
    
    x = Conv2D(16, (3, 3), strides = (2,2), padding = 'same')(model_input)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(16, (3, 3), strides = (2,2), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    xx = Conv2D(16, (1,1), strides= (4,4), padding = 'same')(model_input)
    x = Add()([x,xx])
    xx = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(32, (3, 3), strides = (1,1), padding = 'same')(xx)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(32, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    xx = Conv2D(32, (1,1), strides= (1,1), padding = 'same')(xx)
    x = Add()([x,xx])
    xx = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(48, (3, 3), strides = (1,1), padding = 'same')(xx)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(48, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    xx = Conv2D(48, (1,1), strides= (1,1), padding = 'same')(xx)
    x = Add()([x,xx])
    xx = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(64, (3, 3), strides = (1,1), padding = 'same')(xx)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(64, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    xx = Conv2D(64, (1,1), strides= (1,1), padding = 'same')(xx)
    x = Add()([x,xx])
    xx = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
    
    x = Conv2D(80, (3, 3), strides = (1,1), padding = 'same')(xx)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    x = Conv2D(80, (3, 3), strides = (1,1), padding = 'same')(x)
    x = BatchNormalization()(x)
    x = Activation('relu')(x)
    xx = Conv2D(80, (1,1), strides= (1,1), padding = 'same')(xx)
    x = Add()([x,xx])
    xx = MaxPooling2D(pool_size=(3, 3), strides=(2,2))(x)
            
    x = GlobalAveragePooling2D()(xx)
    x = Dense(num_classes, activation='softmax')(x)
    model = Model(inputs=model_input, outputs=x, name='custom_cnn')
    return model

#instantiate the model
custom_model = custom_cnn(model_input)

#display model summary
custom_model.summary()