ValueError：未知的初始值设定项：HeNormal

如何解决ValueError：未知的初始值设定项：HeNormal

我正在建立一个基于cnn的情绪检测模型。我已经在google colab上训练了模型，并尝试将这些权重加载到本地的jupyter机器中。这样做时，我得到以下错误堆栈：

self.parent.ids.lblA.text = "I am changed from btnA(which is in Table module)"

我的代码部分是：

ValueError                                Traceback (most recent call last)
<ipython-input-9-95fbe8e6b6b2> in <module>
     21 
     22 path = ""
---> 23 model = load_model(path)
     24 
     25 fcc_path = "Tools/haarcascade_frontalface_alt.xml"

<ipython-input-9-95fbe8e6b6b2> in load_model(path)
      9         json_file.close()
     10 
---> 11         model = model_from_json(loaded_model_json)
     12         model.load_weights(path + "model.h5")
     13         print("Loaded model from disk")

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\saving\model_config.py in model_from_json(json_string,custom_objects)
     94   config = json.loads(json_string)
     95   from tensorflow.python.keras.layers import deserialize  # pylint: disable=g-import-not-at-top
---> 96   return deserialize(config,custom_objects=custom_objects)

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\layers\serialization.py in deserialize(config,custom_objects)
    100       module_objects=globs,101       custom_objects=custom_objects,--> 102       printable_module_name='layer')

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\utils\generic_utils.py in deserialize_keras_object(identifier,module_objects,custom_objects,printable_module_name)
    189             custom_objects=dict(
    190                 list(_GLOBAL_CUSTOM_OBJECTS.items()) +
--> 191                 list(custom_objects.items())))
    192       with CustomObjectScope(custom_objects):
    193         return cls.from_config(cls_config)

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\engine\sequential.py in from_config(cls,config,custom_objects)
    367     for layer_config in layer_configs:
    368       layer = layer_module.deserialize(layer_config,--> 369                                        custom_objects=custom_objects)
    370       model.add(layer)
    371     if not model.inputs and build_input_shape:

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\layers\serialization.py in deserialize(config,printable_module_name)
    191                 list(custom_objects.items())))
    192       with CustomObjectScope(custom_objects):
--> 193         return cls.from_config(cls_config)
    194     else:
    195       # Then `cls` may be a function returning a class.

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\engine\base_layer.py in from_config(cls,config)
    592         A layer instance.
    593     """
--> 594     return cls(**config)
    595 
    596   def compute_output_shape(self,input_shape):

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\layers\convolutional.py in __init__(self,filters,kernel_size,strides,padding,data_format,dilation_rate,activation,use_bias,kernel_initializer,bias_initializer,kernel_regularizer,bias_regularizer,activity_regularizer,kernel_constraint,bias_constraint,**kwargs)
    489         activation=activations.get(activation),490         use_bias=use_bias,--> 491         kernel_initializer=initializers.get(kernel_initializer),492         bias_initializer=initializers.get(bias_initializer),493         kernel_regularizer=regularizers.get(kernel_regularizer),~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\initializers.py in get(identifier)
    192     return None
    193   if isinstance(identifier,dict):
--> 194     return deserialize(identifier)
    195   elif isinstance(identifier,six.string_types):
    196     identifier = str(identifier)

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\initializers.py in deserialize(config,custom_objects)
    184       module_objects=module_objects,185       custom_objects=custom_objects,--> 186       printable_module_name='initializer')
    187 
    188 

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\utils\generic_utils.py in deserialize_keras_object(identifier,printable_module_name)
    178     config = identifier
    179     (cls,cls_config) = class_and_config_for_serialized_keras_object(
--> 180         config,printable_module_name)
    181 
    182     if hasattr(cls,'from_config'):

~\Anaconda3\lib\site-packages\tensorflow_core\python\keras\utils\generic_utils.py in class_and_config_for_serialized_keras_object(config,printable_module_name)
    163     cls = module_objects.get(class_name)
    164     if cls is None:
--> 165       raise ValueError('Unknown ' + printable_module_name + ': ' + class_name)
    166   return (cls,config['config'])
    167 

ValueError: Unknown initializer: HeNormal

希望您能找到解决我的问题的方法，并且过去几天一直停留在这里。预先感谢。

解决方法

您的推理环境中的 tensorflow/keras 版本似乎与您用于训练模型的版本不同。他们在较新版本中将 he_normal 更改为 HeNormal。

尝试以下操作，它应该让 tensorflow 引用正确的对象：

HeNormal = tf.keras.initializers.he_normal()
load_model= keras.models.load_model(your_model_path,custom_objects={'HeNormal': HeNormal},compile=False)