Keras之模型拼接

早上起来看到一篇微信公众号推文，看到了新冠肺炎检测的demo，于是阅读起了源代码，原文链接：https://mp.weixin.qq.com/s?__biz=MzUzODkxNzQzMw==&mid=2247484133&idx=1&sn=a4fdd0a6f3a884e94d973f4ca308ab0f&chksm=fad12db3cda6a4a56f3eb5225cea2077bdc9fb11c707434ea0cd80da9906169db187da0ee317&mpshare=1&scene=23&srcid=&sharer_sharetime=1584490235032&sharer_shareid=60d973e8e96477776dbe7e0700964da0#rd

里面有一段模型加载和网络设置的，我不是太懂，代码如下：

为什么既有baseModel，又有headModel呢，于是查阅了相关资料，这里有篇博客讲懂了我，链接：https://www.cnblogs.com/ZhengPeng7/p/9904771.html

说是，在训练较大网络时, 往往想加载预训练的模型, 但若想在网络结构上做些添补, 可能会出现问题，所以需要keras的这种写法。

以单输出回归任务为例：

# 添在末尾:
base_model = InceptionV3(weights='imagenet', include_top=False)
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1, activation='relu')(x)

model = Model(inputs=base_model.input, outputs=x)
model.summary()

# 添在开头和末尾:
# 在开头加1x1卷积层, 使4通道降为3通道, 再传入InceptionV3
def head_model(input_shape=(150, 150, 4)):
    input_tensor = Input(input_shape)
    x = Conv2D(128, (1, 1), activation='relu')(input_tensor)
    x = Conv2D(3, (1, 1), activation='relu')(x)
    model = Model(inputs=input_tensor, outputs=x, name='head')
    return model

head_model = head_model()
body_model = InceptionV3(weights='imagenet', include_top=False)
base_model = Model(head_model.input, body_model(head_model.output))
x = base_model.output
x = GlobalAveragePooling2D()(x)
x = Dense(1, activation='relu')(x)

model = Model(inputs=base_model.inputs, outputs=x, name='net')
base_model.summary()

# 两数据输入流合并于末尾:
base_model = InceptionV3(weights='imagenet', include_top=False, input_shape=(150, 150, 3))
flat = Flatten()(base_model.output)
input_K = Input((100, ))    # another_input
K_flow = Activation(activation='linear')(input_K)
x = concatenate([flat, K_flow])    # 合流
x = Dense(1024, activation='relu')(x)
x = Dense(512, activation='relu')(x)
x = Dense(1, activation='relu')(x)
model = Model(inputs=[*base_model.inputs, input_K], outputs=x)    # 数据生成器那里也以这种形式生成([x_0, x_1], y)即可.
model.summary()