Keras Sequential添加hub.KerasLayer报错，求不换版本的替代方案

尝试将TF Hub上的MobileNetV2模型封装为Keras Layer并加入Sequential模型时，持续报错：Only instances of keras.Layer can be added to a Sequential model，不想更换TensorFlow/Keras版本，寻求替代方案。

使用代码：

mobilenet_v2 ="https://tfhub.dev/google/tf2-preview/mobilenet_v2/classification/4"

classifier_model = mobilenet_v2

IMAGE_SHAPE = (224, 224)

classifier = tf.keras.Sequential([
    hub.KerasLayer(classifier_model, input_shape=IMAGE_SHAPE+(3,))
])

完整报错信息：

ValueError                                Traceback (most recent call last)
Cell In[8], line 3
      1 IMAGE_SHAPE = (224, 224)
----> 3 classifier = tf.keras.Sequential([
      4     hub.KerasLayer(classifier_model, input_shape=IMAGE_SHAPE+(3,))
      5 ])

File ~\Desktop\TFproj\tfvenv\Lib\site-packages\keras\src\models\sequential.py:73, in Sequential.__init__(self, layers, trainable, name)
     71 if layers:
     72     for layer in layers:
---> 73         self.add(layer, rebuild=False)
     74     self._maybe_rebuild()

File ~\Desktop\TFproj\tfvenv\Lib\site-packages\keras\src\models\sequential.py:95, in Sequential.add(self, layer, rebuild)
     93         layer = origin_layer
     94 if not isinstance(layer, Layer):
---> 95     raise ValueError(
     96         "Only instances of `keras.Layer` can be "
     97         f"added to a Sequential model. Received: {layer} "
     98         f"(of type {type(layer)})"
     99     )
    100 if not self._is_layer_name_unique(layer):
    101     raise ValueError(
    102         "All layers added to a Sequential model "
    103         f"should have unique names. Name '{layer.name}' is already "
    104         "the name of a layer in this model. Update the `name` argument "
    105         "to pass a unique name."
    106     )

ValueError: Only instances of `keras.Layer` can be added to a Sequential model. Received: <tensorflow_hub.keras_layer.KerasLayer object at 0x0000021DD8434FB0> (of type <class 'tensorflow_hub.keras_layer.KerasLayer'>)

问题根源是当前环境中tensorflow_hub.KerasLayer与Keras原生Layer类的继承关系不被Sequential模型的类型检查认可，以下两种方法无需更换版本即可解决：

方法1：自定义Keras Layer包装TF Hub模型 #

手动实现一个继承自tf.keras.layers.Layer的包装类，内部调用TF Hub的模型，绕过类型检查：

import tensorflow as tf
import tensorflow_hub as hub

mobilenet_v2 = "https://tfhub.dev/google/tf2-preview/mobilenet_v2/classification/4"
IMAGE_SHAPE = (224, 224)

class HubWrapperLayer(tf.keras.layers.Layer):
    def __init__(self, hub_model_url, **kwargs):
        super().__init__(**kwargs)
        self.hub_layer = hub.KerasLayer(hub_model_url)
    
    def call(self, inputs):
        return self.hub_layer(inputs)

# 构建Sequential模型
classifier = tf.keras.Sequential([
    HubWrapperLayer(mobilenet_v2, input_shape=IMAGE_SHAPE+(3,))
])

方法2：使用函数式API替代Sequential模型 #

直接用Keras函数式API构建模型，避开Sequential的类型检查限制：

import tensorflow as tf
import tensorflow_hub as hub

mobilenet_v2 = "https://tfhub.dev/google/tf2-preview/mobilenet_v2/classification/4"
IMAGE_SHAPE = (224, 224)

inputs = tf.keras.Input(shape=IMAGE_SHAPE+(3,))
x = hub.KerasLayer(mobilenet_v2)(inputs)
classifier = tf.keras.Model(inputs=inputs, outputs=x)

内容的提问来源于stack exchange，提问作者inter galactic