depriciated torch.utils.ffi used

[research-boy]

Traceback (most recent call last):
File "build.py", line 3, in
from torch.utils.ffi import create_extension
File "/usr/local/lib/python3.6/dist-packages/torch/utils/ffi/init.py", line 1, in
raise ImportError("torch.utils.ffi is deprecated. Please use cpp extensions instead.")
ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.
Traceback (most recent call last):
File "build_modulated.py", line 3, in
from torch.utils.ffi import create_extension
File "/usr/local/lib/python3.6/dist-packages/torch/utils/ffi/init.py", line 1, in
raise ImportError("torch.utils.ffi is deprecated. Please use cpp extensions instead.")
ImportError: torch.utils.ffi is deprecated. Please use cpp extensions instead.

[JamesZhan107]

Have you solved this problem？ @research-boy

[research-boy]

@JamesZhan107 Not yet, let me know if you can help me with this

[JamesZhan107]

@JamesZhan107 Not yet, let me know if you can help me with this

https://github.com/ruinmessi/ASFF/tree/09f558863ae297663ec476f52a90c0cf6881415e/utils/DCN
This repo used dcnv2 and has modified the module. It can be complied.

[chengdazhi]

If you guys are using more up to date pytorch, you can checkout the deformable conv module in mmdetection repo.

[JamesZhan107]

‘’‘
inputs = Variable(torch.randn(N, inC, inH, inW).cuda(), requires_grad=True)
offset = conv(inputs)
#offset = Variable(torch.randn(N, num_deformable_groups * 2 * kH * kW, inH, inW).cuda(), requires_grad=True)
output = conv_offset2d(inputs, offset)
‘’’
As is shown in test.py, are these two inputs the same? @chengdazhi

[chengdazhi]

Yep

[JamesZhan107]

Yep

OK, Thanks you !

[herbertcai]

Hi @chengdazhi,

I tried the DCN from here: https://github.com/chengdazhi/Deformable-Convolution-V2-PyTorch/tree/mmdetection/mmdet/ops/dcn. It still gives the depreciation error. I then went to the official mmdetection repo https://github.com/open-mmlab/mmdetection, but couldn't find the DCN module. Could you point me to the right location?

Thanks!