windows10下 centernet导出onnx问题

---

首先demo_det_r101.py已经可以正常跑通，然后我在使用export_det_r101_onnx.py时，报错

Traceback (most recent call last):
  File "E:/LearningCodes/centernet/centernet-master/export_det_r101_onnx.py", line 191, in <module>
    detector.export_onnx(data_f, 'centernet_r101_coco.onnx')
  File "E:/LearningCodes/centernet/centernet-master/export_det_r101_onnx.py", line 179, in export_onnx
    input_names=['image'],
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\onnx\__init__.py", line 25, in export
    return utils.export(*args, **kwargs)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\onnx\utils.py", line 131, in export
    strip_doc_string=strip_doc_string)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\onnx\utils.py", line 363, in _export
    _retain_param_name, do_constant_folding)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\onnx\utils.py", line 266, in _model_to_graph
    graph, torch_out = _trace_and_get_graph_from_model(model, args, training)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\onnx\utils.py", line 225, in _trace_and_get_graph_from_model
    trace, torch_out = torch.jit.get_trace_graph(model, args, _force_outplace=True)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\jit\__init__.py", line 231, in get_trace_graph
    return LegacyTracedModule(f, _force_outplace, return_inputs)(*args, **kwargs)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\nn\modules\module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "D:\Anaconda3\envs\CenterNet\lib\site-packages\torch\jit\__init__.py", line 295, in forward
    out_vars, _ = _flatten(out)
RuntimeError: Only tuples, lists and Variables supported as JIT inputs, but got dict

@Maiya I donnnnn't know which version you were using, but our MANA version already handled this:

@金天 把models/networks/msra_resnet.py中234行

            ret = {}
            for head in self.heads:
                ret[head] = self.__getattr__(head)(x)
            return [ret]

改为

             res = []
             for head in self.heads:
                 res.append(self.__getattr__(head)(x))
             return tuple(res)

就可以正常导出onnx 了，

怎么解决的fedsfsdfsdg

已解决。