InvalidArgumentError (see above for traceback): Input to reshape is a tensor with 401408 values, but the requested shape has 802816

[ZechCal]

10-20 12:02:08 - x - INFO: - Input to reshape is a tensor with 401408 values, but the requested shape has 802816
[[node BatchGather/Reshape_2 (defined at /home/zxy/3dtest/3D-Face-GCNs/utils.py:832) ]]
[[node render_1/strided_slice_34 (defined at /home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/rasterize_triangles.py:121) ]]

Caused by op 'BatchGather/Reshape_2', defined at:
File "main.py", line 142, in
main()
File "main.py", line 95, in main
model = NormalModel(args, sess, graph, refer_mesh, image_paths, img_file)
File "/home/zxy/3dtest/3D-Face-GCNs/model_normal.py", line 16, in init
super(Model, self).init(*args, **kwargs)
File "/home/zxy/3dtest/3D-Face-GCNs/base_model.py", line 115, in init
self.build_graph()
File "/home/zxy/3dtest/3D-Face-GCNs/base_model.py", line 200, in build_graph
pred_results = self.inference(self.train_rgbas, self.coeff, self.image_emb)
File "/home/zxy/3dtest/3D-Face-GCNs/base_model.py", line 354, in inference
proj_color = self.project_color(proj_vert, eros_image)
File "/home/zxy/3dtest/3D-Face-GCNs/base_model.py", line 1029, in project_color
proj_color = utils.batch_gather(flatten_image, coords)
File "/home/zxy/3dtest/3D-Face-GCNs/utils.py", line 718, in batch_gather
return _batch_gather(params, indices, batch_dims=indices.shape.ndims - 1)
File "/home/zxy/3dtest/3D-Face-GCNs/utils.py", line 832, in _batch_gather
flat_params = tf.reshape(params, tf.concat([[flat_inner_shape], outer_shape], axis=0))
File "/home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/tensorflow/python/ops/gen_array_ops.py", line 7179, in reshape
"Reshape", tensor=tensor, shape=shape, name=name)
File "/home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/tensorflow/python/framework/op_def_library.py", line 788, in _apply_op_helper
op_def=op_def)
File "/home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/tensorflow/python/util/deprecation.py", line 507, in new_func
return func(*args, **kwargs)
File "/home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 3300, in create_op
op_def=op_def)
File "/home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/tensorflow/python/framework/ops.py", line 1801, in init
self._traceback = tf_stack.extract_stack()

InvalidArgumentError (see above for traceback): Input to reshape is a tensor with 401408 values, but the requested shape has 802816
[[node BatchGather/Reshape_2 (defined at /home/zxy/3dtest/3D-Face-GCNs/utils.py:832) ]]
[[node render_1/strided_slice_34 (defined at /home/zxy/anaconda3/envs/tensorglue/lib/python3.6/site-packages/rasterize_triangles.py:121) ]]

Several tests proved that the input tensor and requested tensor always have a 401408 difference (no matter how big or small the batch is). Need help......

[ZechCal]

It seems that if change the batch to 2, the problem will be solved...... But why?