How to export onnx model ? RuntimeError: ONNX export failed: Couldn't export Python operator ModulatedDeformConvFunction

[zccyman]

def transformModel2onnx(model_file, inference_file, tensor_size):
    model = torch.load(model_file)
    model = model.cuda()
    
    dummy_input = Variable(torch.randn(tensor_size)).cuda()
    
   #pytorch2onnx
    torch.onnx.export(model, dummy_input, inference_file,
              export_params=True, verbose=True, training=False)

if __name__ == "__main__":
    tensor_size = [1, 7, 3, 64, 128]
    transformModel2onnx(model_file, inference_file, tensor_size)

RuntimeError: ONNX export failed: Couldn't export Python operator ModulatedDeformConvFunction

[shaominghe]

hi ，what has this problem been solved

[xinntao]

Not familiar with ONNX, you may google it ~

[Jeck-Liu-Create]

In order to solve this runtimeerror: onnx export failed: couldn't export python operator modulateddeformconvfunctio problem, you need to make the following adjustments; add the following code to ModulatedDeformConvFunction in dcn/functions/deform_conv.py:

@staticmethod
def symbolic(g, input, offset, mask, weight, bias, stride, padding,
                dilation, groups, deform_groups):
    input_tensors = [input, offset, mask, weight]
    if bias is not None:
        input_tensors.append(bias)
    return g.op(
        'mmcv::MMCVModulatedDeformConv2d',
        *input_tensors,
        stride_i=stride,
        padding_i=padding,
        dilation_i=dilation,
        groups_i=groups,
        deform_groups_i=deform_groups)

The solution comes from: https://zhuanlan.zhihu.com/p/684415514