ONNX can not export Unfold

[Godricly]

🐛 Bug

Unfold can't be exported.
similar issue #22862

To Reproduce

import torch
from torch import nn
import numpy as np

class Demo(torch.nn.Module):
    def __init__(self):
        super().__init__()
        self.unfold = nn.Unfold(kernel_size=3, padding=1)

    def forward(self, x):
        n,c,h,w = x.shape
        unfold_x = self.unfold(x).view(n,-1,h,w)
        return unfold_x

if __name__ == "__main__":
    input_tensor = torch.zeros((1,16,100,100))
    demo = Demo()
    out = demo(input_tensor)
    torch.onnx.export(demo, input_tensor, "debug.onnx", verbose=True,
                        input_names=['data'],
                        opset_version=11,
                        # do_constant_folding=True,
                        dynamic_axes={'data':{0:'batch', 2:'width', 3:'height'},
                        })

Expected behavior

File "/home/godricly/python36/lib/python3.6/site-packages/torch/onnx/symbolic_registry.py", line 91, in get_registered_op
    return _registry[(domain, version)][opname]
KeyError: 'im2col'

Environment

Collecting environment information...
PyTorch version: 1.4.0.dev20191103
Is debug build: No
CUDA used to build PyTorch: 10.1

OS: Ubuntu 18.04.3 LTS
GCC version: (Ubuntu 7.4.0-1ubuntu1~18.04.1) 7.4.0
CMake version: version 3.15.3

Python version: 3.6
Is CUDA available: Yes
CUDA runtime version: 10.1.243
GPU models and configuration: GPU 0: GeForce GTX 1080
Nvidia driver version: 430.50
cuDNN version: /usr/lib/x86_64-linux-gnu/libcudnn.so.7.6.3

Versions of relevant libraries:
[pip3] numpy==1.17.2
[pip3] torch==1.4.0.dev20191103
[pip3] torchvision==0.4.2
[conda] Could not collect

cc @houseroad @spandantiwari @lara-hdr @BowenBao @neginraoof

[mruberry]

@houseroad

[spandantiwari]

@Godricly - there's at least partial support for unfold that went in with this PR #24970. Full support is on the backlog, but we haven't got a chance to get to it yet.
Could you share if you are using im2col in a real model? It will help us prioritize.

[Godricly]

It is used in CARAFE: Content-Aware ReAssembly of FEatures to extract local patches.

[isagastiberri]

I've also run into this same problem:

RuntimeError: ONNX export failed: Couldn't export Python operator Im2Col

I would like to export SiamMask to ONNX. Is there any way to do it in the meantime?

[Godricly]

@spandantiwari Can you make sure that demo code works? Thx

[neginraoof]

PR #30972 addresses this issue.

[GOBish]

@isagastiberri were you able to solve this?