cpu-only install

[korin-lf]

Thank you. I followed 4-7 from (https://github.com/MVIG-SJTU/AlphaPose/issues/817#issuecomment-840260684) and got it working on ubuntu 20.04. Now I am on ubuntu 18.04 with no Nvidia card I followed exactly all 2-8 instructions python36, removed all cuda and installed cuda 10.1 and pytorch==1.7.1 torchvision==0.8.2 cpuonly -c pytorch

removed ext_modules=get_ext_modules(), from setup.py as I am on cpu only machine

Now I am getting : File "/home/user/gesture_detect/gesture_detect/models/layers/dcn/deform_conv.py", line 11, in from . import deform_conv_cuda ImportError: libcudart.so.10.0: cannot open shared object file: No such file or directory

what is the possible way to resolve this? I tried with symlinks but no luck thank you

[divastar]

All the problems : stack on inferense, undefined symbol of deform_conv_cuda just spent my day with nothing on all systems ubuntu 20.04 etc Can someone clarify this install mess? cuda 10.1 with what (pytorch 1.1?) gonna work?

[korin-lf]

Finally got it working on ubuntu 20.04 uninstalled all previous cuda, pytorch started with python3.8 cuda 11.2 drive 460 gcc9.3.0 pytorch 1.8

[P3RI9]

Hi @korin-lf Thanks for your answer! Im meeting the same problem trying to install alphapose on a non-cpu machine. I have some questions:

1. Did you install cpuonly pytorch version for 1.8.1?
2. Did you merge this pull request git pull origin pull/592/head as its said https://github.com/MVIG-SJTU/AlphaPose/blob/master/docs/INSTALL.md?
3. Did you comment get_ext_modules() line?
4. Did you use --gpus -1 when launching demo_inference?

Thanks for your reply!

[lofyol]

Finally got it working on ubuntu 20.04 uninstalled all previous cuda, pytorch started with python3.8 cuda 11.2 drive 460 gcc9.3.0 pytorch 1.8

How do you compile successfully on the high version CUDA

[divastar]

You just have to keep everything compatible all the versions of cuda Python compiler etc

On Wed, 14 Jul 2021 at 4:15 CCxiaoLL @.***> wrote:

Finally got it working on ubuntu 20.04 uninstalled all previous cuda, pytorch started with python3.8 cuda 11.2 drive 460 gcc9.3.0 pytorch 1.8

How do you compile successfully on the high version CUDA

— You are receiving this because you commented. Reply to this email directly, view it on GitHub https://github.com/MVIG-SJTU/AlphaPose/issues/863#issuecomment-879507397, or unsubscribe https://github.com/notifications/unsubscribe-auth/AMB4M4G2S73CALRF7OMFP73TXTQKBANCNFSM45HKO5VQ .

[jeremy-rico]

Having same issue with python 3.6 pytorch 1.1.0 torchvision 0.3.0

I've tried updating all these packages to different versions but nothing fixes. Also commented out line 211 in setup.py and set force_compile to True. Still getting: ImportError: cannot import name 'deform_conv_cuda'

any more info on how you solved this would be much appreciated