search

MVIG-SJTU / AlphaPose

Real-Time and Accurate Full-Body Multi-Person Pose Estimation&Tracking System
http://mvig.org/research/alphapose.html
Other
8.05k stars 1.98k forks source link

3D Inference Script getting stuck on rendering the final images #1172

Open RohaanA opened 1 year ago

RohaanA commented 1 year ago

Hi,

After several headaches I was finally able to run 3D inference script. However, its getting stuck on the last part of the script

image

Can someone please tell me if the ffmpeg errors above the final line are causing this? and if so how can I fix them

hsmwfgie commented 1 year ago

Hi, I am also trying to get the 3D Inference Script to work. But I fail with different shape sizes. Could you possibly post your script call and the used parameters?

RohaanA commented 1 year ago

Hi, I am also trying to get the 3D Inference Script to work. But I fail with different shape sizes. Could you possibly post your script call and the used parameters?

Hey, I can give u the colab notebook I am using, maybe u can figure out the problem. (notebook here)

jmz3 commented 2 weeks ago

Hi, I have the same issue with demo_3d_inference.py. I have tried to probe the issue and now I'm pretty sure it's because of the incompatibility between pytorch3d and pytorch. When I comment out this line in alphapose/utils/vis.py, and this line, the program will be able to proceed and save all the data, but otherwise it will get stuck at the same place you mentioned. Unfortunately, since those lines are for visualization, commenting them out will disable the keypoint and skeleton overlay.

I haven't figured out how to resolve this. I suppose this is a known issue for the developers so I really appreciate it if they can come and help us. Nevertheless, I will try to troubleshoot it a bit further myself and I will post it here if I find anything helpful.