导出onnx报错

[jyh11111]

python tools/deployment/export_onnx.py -c <配置文件路径> -r <检查点路径> [--check] [--simplify] 这个报错，提示如下：

_D:\deim\DEIM> python tools/deployment/export_onnx.py -c configs\deim_dfine\my_deim_hgnetv2_n_coco.yml -r deim_outputs\deim_hgnetv2_n_coco\best_stg2.pth
D:\deim\DEIM\tools\deployment../..\engine\deim\dfine_decoder.py:644: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
if memory.shape[0] > 1:
D:\deim\DEIM\tools\deployment../..\engine\deim\dfine_decoder.py:129: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
if reference_points.shape[-1] == 2:
D:\deim\DEIM\tools\deployment../..\engine\deim\dfine_decoder.py:133: TracerWarning: Converting a tensor to a Python boolean might cause the trace to be incorrect. We can't record the data flow of Python values, so this value will be treated as a constant in the future. This means that the trace might not generalize to other inputs!
elif reference_points.shape[-1] == 4:
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx_internal\jit_utils.py:307: UserWarning: Constant folding - Only steps=1 can be constant folded for opset >= 10 onnx::Slice op. Constant folding not applied. (Triggered internally at C:\cb\pytorch_1000000000000\work\torch\csrc\jit\passes\onnx\constant_fold.cpp:181.)
_C._jit_pass_onnx_node_shape_type_inference(node, params_dict, opset_version)
Traceback (most recent call last):
File "D:\deim\DEIM\tools\deployment\export_onnx.py", line 103, in
main(args)
File "D:\deim\DEIM\tools\deployment\export_onnx.py", line 65, in main
torch.onnx.export(
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py", line 516, in export
_export(
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py", line 1596, in _export
graph, params_dict, torch_out = _model_to_graph(
^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py", line 1139, in _model_to_graph
graph = _optimize_graph(
^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py", line 677, in _optimize_graph
graph = _C._jit_pass_onnx(graph, operator_export_type)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py", line 1940, in _run_symbolic_function
return symbolic_fn(graph_context, *inputs, **attrs)
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_helper.py", line 395, in wrapper
return fn(g, *args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_helper.py", line 306, in wrapper
return fn(g, args, **kwargs)
^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_opset10.py", line 203, in symbolic_fn
padding_ceil = opset9.get_pool_ceil_padding(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_opset9.py", line 1565, in get_pool_ceil_padding
return symbolic_helper._unimplemented(
^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_helper.py", line 612, in _unimplemented
_onnx_unsupported(f"{op}, {msg}", value)
File "D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\symbolic_helper.py", line 623, in _onnx_unsupported
raise errors.SymbolicValueError(
torch.onnx.errors.SymbolicValueError: Unsupported: ONNX export of operator get_pool_ceil_padding, input size not accessible. Please feel free to request support or submit a pull request on PyTorch GitHub: https://github.com/pytorch/pytorch/issues [Caused by the value 'input.8 defined in (%input.8 : Float(, *, *, *, strides=[1648656, 103041, 321, 1], requires_grad=1, device=cpu) = onnx::Pad[mode="constant"](%519, %545, %506), scope: main.main..Model::/engine.deim.deim.DEIM::model/engine.backbone.hgnetv2.HGNetv2::backbone/engine.backbone.hgnetv2.StemBlock::stem # D:\deim\DEIM\tools\deployment../..\engine\backbone\hgnetv2.py:168:0
)' (type 'Tensor') in the TorchScript graph. The containing node has kind 'onnx::Pad'.]
(node defined in D:\deim\DEIM\tools\deployment../..\engine\backbone\hgnetv2.py(168): forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1508): _slow_forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1527): _call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1518): _wrapped_call_impl
D:\deim\DEIM\tools\deployment../..\engine\backbone\hgnetv2.py(537): forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1508): _slow_forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1527): _call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1518): _wrapped_call_impl
D:\deim\DEIM\tools\deployment../..\engine\deim\deim.py(27): forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1508): _slow_forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1527): _call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1518): _wrapped_call_impl
D:\deim\DEIM\tools\deployment\export_onnx.py(48): forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1508): _slow_forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1527): _call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1518): _wrapped_call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\jit_trace.py(124): wrapper
D:\miniconda3\envs\deim\Lib\site-packages\torch\jit_trace.py(133): forward
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1527): _call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\nn\modules\module.py(1518): _wrapped_call_impl
D:\miniconda3\envs\deim\Lib\site-packages\torch\jit_trace.py(1285): _get_trace_graph
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py(915): _trace_and_get_graph_from_model
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py(1011): _create_jit_graph
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py(1135): _model_to_graph
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py(1596): _export
D:\miniconda3\envs\deim\Lib\site-packages\torch\onnx\utils.py(516): export
D:\deim\DEIM\tools\deployment\export_onnx.py(65): main
D:\deim\DEIM\tools\deployment\export_onnx.py(103):
)

Inputs:
    #0: 519 defined in (%519 : Float(*, 16, 320, 320, strides=[1638400, 102400, 320, 1], requires_grad=1, device=cpu) = onnx::Add(%518, %model.backbone.stem.stem1.lab.bias), scope: __main__.main.<locals>.Model::/engine.deim.deim.DEIM::model/engine.backbone.hgnetv2.HGNetv2::backbone/engine.backbone.hgnetv2.StemBlock::stem/engine.backbone.hgnetv2.ConvBNAct::stem1/engine.backbone.hgnetv2.LearnableAffineBlock::lab # D:\deim\DEIM\tools\deployment\../..\engine\backbone\hgnetv2.py:35:0
)  (type 'Tensor')
    #1: 545 defined in (%545 : Long(8, strides=[1], device=cpu) = onnx::Cast[to=7](%544), scope: __main__.main.<locals>.Model::/engine.deim.deim.DEIM::model/engine.backbone.hgnetv2.HGNetv2::backbone/engine.backbone.hgnetv2.StemBlock::stem # D:\deim\DEIM\tools\deployment\../..\engine\backbone\hgnetv2.py:168:0
)  (type 'Tensor')
    #2: 506 defined in (%506 : NoneType = prim::Constant(), scope: __main__.main.<locals>.Model::/engine.deim.deim.DEIM::model/engine.backbone.hgnetv2.HGNetv2::backbone/engine.backbone.hgnetv2.StemBlock::stem/engine.backbone.hgnetv2.ConvBNAct::stem1/torch.nn.modules.conv.Conv2d::conv
)  (type 'NoneType')
Outputs:
    #0: input.8 defined in (%input.8 : Float(*, *, *, *, strides=[1648656, 103041, 321, 1], requires_grad=1, device=cpu) = onnx::Pad[mode="constant"](%519, %545, %506), scope: __main__.main.<locals>.Model::/engine.deim.deim.DEIM::model/engine.backbone.hgnetv2.HGNetv2::backbone/engine.backbone.hgnetv2.StemBlock::stem # D:\deim\DEIM\tools\deployment\../..\engine\backbone\hgnetv2.py:168:0
)  (type 'Tensor')_