Cannot take the length of shape with unknown rank.

[lzfff12]

Describe the bug

Traceback (most recent call last):
File "convert2onnx/convert_superglue_to_onnx.py", line 124, in
main()
File "convert2onnx/convert_superglue_to_onnx.py", line 117, in main
tf_rep.run((kpts0, scores0, desc0, kpts1, scores1, desc1))
File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/backend_rep.py", line 107, in run
output_values = self.tf_module(**input_dict)
File "/mnt/data/anaconda3/envs/lightglue/lib/python3.8/site-packages/tensorflow/python/util/traceback_utils.py", line 153, in error_handler
raise e.with_traceback(filtered_tb) from None
File "/tmp/autograph_generated_filecuhrqz0v.py", line 30, in tf____call
ag__.for_stmt(ag__.ld(self).graph_def.node, None, loop_body, get_state, set_state, (), {'iterate_names': 'node'})
File "/tmp/autograph_generated_filecuhrqz0v.py", line 23, in loop_body
output_ops = ag.converted_call(ag__.ld(self).backend.onnx_node_to_tensorflow_op, (ag_.ld(onnx_node), ag__.ld(tensor_dict), ag__.ld(self).handlers), dict(opset=ag__.ld(self).opset, strict=ag__.ld(self).strict), fscope)
File "/tmp/autograph_generated_file05s8tr06.py", line 50, in tf___onnx_node_to_tensorflow_op
ag.if_stmt(ag__.ld(handlers), if_body_1, else_body_1, get_state_1, set_state_1, ('do_return', 'retval_'), 2)
File "/tmp/autograph_generated_file05s8tr06.py", line 44, in if_body_1
ag.if_stmt(ag__.ld(handler), if_body, else_body, get_state, set_state, ('do_return', 'retval_'), 2)
File "/tmp/autograph_generated_file05s8tr06.py", line 36, in if_body
retval = ag_.converted_call(ag__.ld(handler).handle, (ag__.ld(node),), dict(tensor_dict=ag__.ld(tensor_dict), strict=ag__.ld(strict)), fscope)
File "/tmp/autograph_generated_filegv1w6atb.py", line 34, in tf__handle
ag.if_stmt(ag__.ld(ver_handle), if_body, else_body, get_state, set_state, ('do_return', 'retval_'), 2)
File "/tmp/autograph_generated_filegv1w6atb.py", line 26, in if_body
retval = ag_.converted_call(ag__.ld(ver_handle), (ag__.ld(node),), dict(**ag__.ld(kwargs)), fscope)
File "/tmp/autograph_generated_fileg_gf0gau.py", line 12, in tf__version
retval = ag_.converted_call(ag__.ld(cls).conv, (ag__.ld(node), ag__.ld(kwargs)['tensor_dict']), None, fscope)
File "/tmp/autograph_generated_filekbecj_e4.py", line 12, in tf__conv
x_rank = ag.converted_call(ag__.ld(len), (ag__.converted_call(ag__.ld(x).get_shape, (), None, fscope),), None, fscope)
ValueError: in user code:

File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/backend_tf_module.py", line 99, in __call__  *
    output_ops = self.backend._onnx_node_to_tensorflow_op(onnx_node,
File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/backend.py", line 347, in _onnx_node_to_tensorflow_op  *
    return handler.handle(node, tensor_dict=tensor_dict, strict=strict)
File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/handlers/handler.py", line 59, in handle  *
    return ver_handle(node, **kwargs)
File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/handlers/backend/conv.py", line 15, in version_11  *
    return cls.conv(node, kwargs["tensor_dict"])
File "/mnt/training/map/zhifeng.lin/code/SuperPoint-SuperGlue-TensorRT/onnx-tensorflow/onnx_tf/handlers/backend/conv_mixin.py", line 30, in conv  *
    x_rank = len(x.get_shape())

ValueError: Cannot take the length of shape with unknown rank.

To Reproduce

torch.onnx.export(superpoint_model, # model being run
(kpts0, scores0, desc0, kpts1, scores1, desc1), # model input (or a tuple for multiple inputs)
onnx_filename, # where to save the model (can be a file or file-like object)
#export_params=True, # store the trained parameter weights inside the model file
opset_version=13, # the ONNX version to export the model to
#do_constant_folding=True, # whether to execute constant folding for optimization
input_names=["keypoints_0", # batch x feature_number0 x 2
"scores_0", # batch x feature_number0
"descriptors_0", # batch x feature_dims x feature_number0
"keypoints_1", # batch x feature_number1 x 2
"scores_1", # batch x feature_number1
"descriptors_1", # batch x feature_dims x feature_number1
], # the model input names
output_names=["scores"], # the model output names
dynamic_axes={'keypoints_0': {1: 'feature_number_0'},
'scores_0': {1: 'feature_number_0'},
'descriptors_0': {2: 'feature_number_0'},
'keypoints_1': {1: 'feature_number_1'},
'scores_1': {1: 'feature_number_1'},
'descriptors_1': {2: 'feature_number_1'},
} # dynamic model input names
)

# check onnx model
onnx_model = onnx.load(onnx_filename)
onnx.checker.check_model(onnx_model)

print("Exported model has been checked with ONNXRuntime.")


#spsg_model_path = "sgsg.onnx"
sgsg_tf_model="superglue.tf.pb"
#model_onnx = onnx.load(spsg_model_path)
tf_rep = prepare(onnx_model)

# Input nodes to the model
print('inputs:', tf_rep.inputs)

Output nodes from the model

print('outputs:', tf_rep.outputs)

All nodes in the model

print('tensor_dict:')
print(tf_rep.tensor_dict)
print("11111")
tf_rep.run((kpts0, scores0, desc0, kpts1, scores1, desc1))
print("2222")
# Export model as .pb file
tf_rep.export_graph(sgsg_tf_model)

A self-contained piece of code that can demonstrate the problem is required.

Please do not expect us to have PyTorch, Caffe2 installed.

If a model exported from PyTorch and Caffe2 is having trouble in ONNX-TF, use the next section to attach the model.

ONNX model file

If applicable, attach the onnx model file in question using Gist, DropBox or Google Drive.

ONNX-TF

This section can be obtained by running get_version.py from util folder.

• Python 3.8:
• ONNX version 1.16.1:
• ONNX-TF version 1.10.1:
• Tensorflow version 2.13.1:

Additional context

Add any other context about the problem here.