ffi import error

[JuChanSeo]

Does anyone succeed run trainval_net_global_local.py?

I have tried to run the code, but there is an import error on pytorch ffi module. Below is error message.

File "/home/.../cddod-master/lib/model/roi_crop/_ext/roi_crop/init.py", line 3, in from ._roi_crop import lib as _lib, ffi as _ffi ImportError: /home/.../cddod-master/lib/model/roi_crop/_ext/roi_crop/_roi_crop.so: undefined symbol: __cudaPopCallConfiguration

I think there is a problem with cuda version and pytorch version. I wonder if there was a problem like this for those who succeeded in running trainval_net_global_local.py, and if so, how did you guys solved it.

Below is my environment

-ubuntu 18.04

• Python 3.6.2
• torch0.4.1 / torch vision 0.2.1
• cuda 10.1
• cudatoolkit 10.1

Thank you.

[Howardqlz]

i have met the same problem. This repo depend on this repo, it seems like we should set an environment like torch 0.4.0, cuda 8.0. I use cuda 9.0 still met this problem.