Question

我正在尝试使用我自己的数据集运行 Tensorflow 教程 https://www.tensorflow.org/tutorials/generative/dcgan 中给出的 DCGAN 代码。要将数据批量加载到内存中，我使用以下代码：

def get_image(image_path):
    image = Image.open(image_path)
    return np.array(np.asarray(image))/255.0

def get_batch(image_files):
    data_batch = np.array(
        [get_image("DEBUG/real/"+sample_file) for sample_file in image_files]).astype(np.float32)
    if len(data_batch.shape) < 4:
        data_batch = data_batch.reshape(data_batch.shape + (1,))
    return data_batch

def get_batches(batch_size):
    current_index = 0
    while current_index + batch_size <= shape[0]:
        data_batch = get_batch(data_files[current_index:current_index + batch_size])
        current_index += batch_size
        yield data_batch / 255.0

与教程的唯一区别是我的输入尺寸为 640 x 360 x 3。我在运行时收到以下警告和错误：

WARNING:tensorflow:Model was constructed with shape Tensor("conv2d_11_input:0", shape=(None, 640, 360, 3), dtype=float32) for input (None, 640, 360, 3), but it was re-called on a Tensor with incompatible shape (0, 1).
ValueError: Shapes (0, 1) and (None, None, None, None) must have the same rank
<ipython-input-92-b17a80c6bb7b> in train_step(images)
      6         generated_images = generator(noise, training=True)
      7 
----> 8         real_output = discriminator(images, training=True)
      9         fake_output = discriminator(generated_images, training=True)
     10 

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\engine\base_layer.py in __call__(self, inputs, *args, **kwargs)
    820           with base_layer_utils.autocast_context_manager(
    821               self._compute_dtype):
--> 822             outputs = self.call(cast_inputs, *args, **kwargs)
    823           self._handle_activity_regularization(inputs, outputs)
    824           self._set_mask_metadata(inputs, outputs, input_masks)

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\engine\sequential.py in call(self, inputs, training, mask)
    265       if not self.built:
    266         self._init_graph_network(self.inputs, self.outputs, name=self.name)
--> 267       return super(Sequential, self).call(inputs, training=training, mask=mask)
    268 
    269     outputs = inputs  # handle the corner case where self.layers is empty

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\engine\network.py in call(self, inputs, training, mask)
    715     return self._run_internal_graph(
    716         inputs, training=training, mask=mask,
--> 717         convert_kwargs_to_constants=base_layer_utils.call_context().saving)
    718 
    719   def compute_output_shape(self, input_shape):

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\engine\network.py in _run_internal_graph(self, inputs, training, mask, convert_kwargs_to_constants)
    889 
    890           # Compute outputs.
--> 891           output_tensors = layer(computed_tensors, **kwargs)
    892 
    893           # Update tensor_dict.

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\engine\base_layer.py in __call__(self, inputs, *args, **kwargs)
    820           with base_layer_utils.autocast_context_manager(
    821               self._compute_dtype):
--> 822             outputs = self.call(cast_inputs, *args, **kwargs)
    823           self._handle_activity_regularization(inputs, outputs)
    824           self._set_mask_metadata(inputs, outputs, input_masks)

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\keras\layers\convolutional.py in call(self, inputs)
    201           strides=self.strides,
    202           padding=self._padding_op,
--> 203           data_format=self._conv_op_data_format)
    204 
    205     # Apply causal padding to inputs for Conv1D.

~\miniconda3\envs\newenv\lib\site-packages\tensorflow_core\python\ops\nn_ops.py in __init__(self, input_shape, filter_shape, padding, strides, dilation_rate, name, data_format)
   1066     except ValueError:
   1067       raise ValueError(
-> 1068           "input tensor must have rank %d" % (num_spatial_dims + 2))
   1069 
   1070     try:

ValueError: input tensor must have rank 4

我没有得到的是形状为 (0,1) 的张量来自哪里？请帮我。我的非常重要的项目需要它，使用 GAN 进行面部活体检测。谢谢。

模型是用形状张量构造的，但它在形状不兼容的张量上被重新调用。使用 Tensorflow DCGAN

0 个答案: