Python tf.random\u crop不能与通配符一起使用？

我正在用Python3.5+TensorFlow1.8构建一个图像翻译网络。 对于数据扩充，我尝试使用带有通配符的

tf.random_crop（）

，如下所示：

# input images
A = tf.placeholder(tf.float, shape=(None, 480, 640, 3))
B = tf.placeholder(tf.float, shape=(None, 480, 640, 3))

# images concatenation to crop on the same random seed
AB = tf.concat([A, B], 3)

# random cropping with wildcard for batch_size specification
AB_cropped = tf.random_crop(AB, [-1, 480, 480, 4])

# cropped images
A_ = AB_cropped[:,:,:,:3]
B_ = AB_cropped[:,:,:,3:]

...

每次运行都会出现一些不同的错误（有时会产生错误的结果）。 发生的错误如下所示：

# input images
A = tf.placeholder(tf.float, shape=(None, 480, 640, 3))
B = tf.placeholder(tf.float, shape=(None, 480, 640, 3))

# images concatenation to crop on the same random seed
AB = tf.concat([A, B], 3)

# random cropping with wildcard for batch_size specification
AB_cropped = tf.random_crop(AB, [-1, 480, 480, 4])

# cropped images
A_ = AB_cropped[:,:,:,:3]
B_ = AB_cropped[:,:,:,3:]

...

InvalidArgumentError（回溯请参见上文）：应在[0,1]中开始[0]，但得到2

回溯（最近一次呼叫最后一次）：
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1350行，在
返回fn（*args）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1329行，在
状态，运行（元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/framework/errors\u impl.py”，第473行，在退出中__
c_api.TF_GetCode（self.status.status））
tensorflow.python.framework.errors\u impl.InvalidArgumentError:应在[0,1]中开始[0]，但得到2
[[Node:preprocess/random_crop=Slice[Index=DT_INT32，T=DT_FLOAT，_device=“/job:localhost/replica:0/task:0/device:CPU:0”]（预处理/concat，预处理/random_crop/mod，预处理/random_crop/size）]]
在处理上述异常期间，发生了另一个异常：
回溯（最近一次呼叫最后一次）：
文件“alpha_gan_-based.py”，第247行，在
_例如，sess.run（[hybrid_op，hybrid_loss]，{image:image_batch，depth:depth_batch，z_prior:sample_z（）}）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第895行，正在运行
运行_元数据_ptr）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1128行，正在运行
feed_dict_tensor、options、run_元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1344行，运行
选项，运行（元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1363行，在_do_调用中
提升类型（e）（节点定义、操作、消息）
tensorflow.python.framework.errors\u impl.InvalidArgumentError:应在[0,1]中开始[0]，但得到2
[[Node:preprocess/random_crop=Slice[Index=DT_INT32，T=DT_FLOAT，_device=“/job:localhost/replica:0/task:0/device:CPU:0”]（预处理/concat，预处理/random_crop/mod，预处理/random_crop/size）]]
由op“预处理/随机裁剪”引起，定义于：
文件“alpha_gan_-based.py”，第156行，在
裁剪=（tf.random_裁剪（合并，[-1191924]）/255）*2-1
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/ops/random_ops.py”，第316行，随机裁剪
返回数组_ops.slice（值、偏移量、大小、名称=名称）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/ops/array_ops.py”，第625行，在切片中
返回gen\u数组操作。\u切片（输入、开始、大小、名称=名称）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/ops/gen_array_ops.py”，第4687行，在
“切片”，输入=输入，开始=开始，大小=大小，名称=名称）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/framework/op_def_library.py”，第787行，位于“应用”和“操作”帮助程序中
op_def=op_def）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/framework/ops.py”，第3160行，在create_op中
op_def=op_def）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/framework/ops.py”，第1625行，在__
self._traceback=self._graph._extract_stack（）35; pylint:disable=protected access
InvalidArgumentError（回溯请参见上文）：应在[0,1]中开始[0]，但得到2
[[Node:preprocess/random_crop=Slice[Index=DT_INT32，T=DT_FLOAT，_device=“/job:localhost/replica:0/task:0/device:CPU:0”]（预处理/concat，预处理/random_crop/mod，预处理/random_crop/size）]]

“重塑”无法推断空张量的缺失输入大小，除非所有指定的输入大小均为非零

回溯（最近一次呼叫最后一次）：
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1350行，在
返回fn（*args）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1329行，在
状态，运行（元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/framework/errors\u impl.py”，第473行，在退出中__
c_api.TF_GetCode（self.status.status））
tensorflow.python.framework.errors\u impl.InvalidArgumentError:除非所有指定的输入大小均为非零，否则重塑无法推断空张量缺少的输入大小
[[Node:encoder/Flatte/Reforme=Reforme[T=DT_FLOAT，Tshape=DT_INT32，_device=“/job:localhost/replica:0/task:0/device:CPU:0”]（encoder/Relu_6，encoder/Flatte/Reforme/shape）]]
在处理上述异常期间，发生了另一个异常：
回溯（最近一次呼叫最后一次）：
文件“alpha_ganu-based.py”，第248行，在
_例如，sess.run（[hybrid_op，hybrid_loss]，{image:image_batch，depth:depth_batch，z_prior:sample_z（）}）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第895行，正在运行
运行_元数据_ptr）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1128行，正在运行
feed_dict_tensor、options、run_元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1344行，运行
选项，运行（元数据）
文件“/usr/tensorflow/lib/python3.5/site packages/tensorflow/python/client/session.py”，第1363行，在_do_调用中
提升类型（e）（节点定义、操作、消息）
tensorflow.python.framework.errors\u impl.InvalidArgumentError:除非所有指定的输入大小均为非零，否则重塑无法推断空张量缺少的输入大小
[[Node:encoder/plant/restrape=restrape[T=DT_FLOAT，Tshape=DT_INT32，_device=“/job:localhost/replica:0/task:
Traceback (most recent call last):
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1350, in _do_call
         return fn(*args)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1329, in _run_fn
         status, run_metadata)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/framework/errors_impl.py", line 473, in __exit__
         c_api.TF_GetCode(self.status.status))
     tensorflow.python.framework.errors_impl.InvalidArgumentError: Reshape cannot infer the missing input size for an empty tensor unless all specified input sizes are non-zero
         [[Node: encoder/flatten/Reshape = Reshape[T=DT_FLOAT, Tshape=DT_INT32, _device="/job:localhost/replica:0/task:0/device:CPU:0"](encoder/Relu_6, encoder/flatten/Reshape/shape)]]

     During handling of the above exception, another exception occurred:

     Traceback (most recent call last):
       File "alpha_gan_based.py", line 248, in <module>
         _, eg_loss = sess.run([hybrid_op, hybrid_loss], {image:image_batch, depth:depth_batch, z_prior:sample_z()})
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 895, in run
         run_metadata_ptr)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1128, in _run
         feed_dict_tensor, options, run_metadata)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1344, in _do_run
         options, run_metadata)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/client/session.py", line 1363, in _do_call
         raise type(e)(node_def, op, message)
     tensorflow.python.framework.errors_impl.InvalidArgumentError: Reshape cannot infer the missing input size for an empty tensor unless all specified input sizes are non-zero
         [[Node: encoder/flatten/Reshape = Reshape[T=DT_FLOAT, Tshape=DT_INT32, _device="/job:localhost/replica:0/task:0/device:CPU:0"](encoder/Relu_6, encoder/flatten/Reshape/shape)]]

     Caused by op 'encoder/flatten/Reshape', defined at:
       File "alpha_gan_based.py", line 163, in <module>
         z_encoded, intermidiate = encoder(x_real_image)
       File "alpha_gan_based.py", line 54, in encoder
         x = tf.layers.flatten(x)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/layers/core.py", line 414, in flatten
         return layer.apply(inputs)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/layers/base.py", line 762, in apply
         return self.__call__(inputs, *args, **kwargs)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/layers/base.py", line 652, in __call__
         outputs = self.call(inputs, *args, **kwargs)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/layers/core.py", line 376, in call
         outputs = array_ops.reshape(inputs, (array_ops.shape(inputs)[0], -1))
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/ops/gen_array_ops.py", line 3997, in reshape
         "Reshape", tensor=tensor, shape=shape, name=name)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/framework/op_def_library.py", line 787, in _apply_op_helper
         op_def=op_def)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/framework/ops.py", line 3160, in create_op
         op_def=op_def)
       File "/usr/tensorflow/lib/python3.5/site-packages/tensorflow/python/framework/ops.py", line 1625, in __init__
         self._traceback = self._graph._extract_stack()  # pylint: disable=protected-access

     InvalidArgumentError (see above for traceback): Reshape cannot infer the missing input size for an empty tensor unless all specified input sizes are non-zero
         [[Node: encoder/flatten/Reshape = Reshape[T=DT_FLOAT, Tshape=DT_INT32, _device="/job:localhost/replica:0/task:0/device:CPU:0"](encoder/Relu_6, encoder/flatten/Reshape/shape)]]