加载Keras的VGG16模型时Python内核崩溃，求助解决

问题场景 #

执行以下代码时出现Python内核崩溃：

model = vgg16.VGG16(input_tensor=input_img, weights='imagenet', include_top=False)

错误信息 #

WARNING:tensorflow:The following Variables were used in a Lambda layer's call (tf.concat), but are not present in its tracked objects: <tf.Variable 'Variable:0' shape=(1, 474, 712, 3) dtype=float32>
<tf.Variable 'Variable:0' shape=(1, 474, 712, 3) dtype=float32>. This is a strong indication that the Lambda layer should be rewritten as a subclassed Layer.
Canceled future for execute_request message before replies were done
The Kernel crashed while executing code in the the current cell or a previous cell. Please review the code in the cell(s) to identify a possible cause of the failure. Click here for more info. View Jupyter log for further details.

排查与解决步骤 #

• 修正输入张量类型：报错核心是Lambda层引用了未被追踪的Variable，大概率是input_img为tf.Variable类型，而VGG16要求input_tensor传入普通张量。将其转换为张量：

  input_img = tf.convert_to_tensor(input_img)
  

  或直接用tf.keras.Input定义输入：

  input_img = tf.keras.Input(shape=(474, 712, 3))
  

• 修复Lambda层变量追踪问题：若代码中自定义了Lambda层并引用外部Variable，需让层正确追踪变量：

  • 改用子类化Layer替代Lambda层：

    class ConcatLayer(tf.keras.layers.Layer):
        def __init__(self, external_var, **kwargs):
            super().__init__(**kwargs)
            self.external_var = external_var
        
        def call(self, inputs):
            return tf.concat([inputs, self.external_var], axis=-1)
    
    x = ConcatLayer(external_var)(input)
    

  • 或在Lambda层中显式传递变量：

    x = tf.keras.layers.Lambda(lambda x, var: tf.concat([x, var], axis=-1),
                               arguments={'var': external_var})(input)
    

• 清理会话并重启内核：旧会话残留可能导致变量追踪混乱，先清理再重启：

  tf.keras.backend.clear_session()
  

  重启Jupyter内核后重新运行代码。

• 检查版本兼容性：确保TensorFlow与Keras版本匹配，建议使用TensorFlow 2.x最新稳定版，避免版本不兼容引发底层错误。

内容的提问来源于stack exchange，提问作者Ignacio