Skip to content

After updating to the newest version, segmentation fault (core dumped) occurs in the section of pose_estimation (multiHMR_896_L.pt) #71

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
senbinyu opened this issue Apr 3, 2025 · 10 comments
Assignees

Comments

@senbinyu
Copy link

senbinyu commented Apr 3, 2025

Thanks for your great work. I followed the instructions of install_cu118.sh in the readme exactly and installed all the dependencies. There is a sementation fault error occured in the self.pose_estimator(image_path) function. Sepecifically, it comes from pose_estimator.py, line 138, self.mhmr_model(). I double checked it in another clean file (only loading the model and inference) and the error occured again. By the way, the model dict can be loaded correctly, the error occurs in the forward process.

@lingtengqiu
Copy link
Collaborator

Do you install the mmcv?

@lingtengqiu
Copy link
Collaborator

@liphao99

@senbinyu
Copy link
Author

senbinyu commented Apr 3, 2025

Yes, I installed mmcv==1.3.9 and ViTPose, ultralytics as extra dependencies for video motion processing. I ran inference.sh and inference_mesh.sh without any change, both of them failed.

@lingtengqiu
Copy link
Collaborator

what is your error logger?

@lingtengqiu
Copy link
Collaborator

What devices you used?

@senbinyu
Copy link
Author

senbinyu commented Apr 3, 2025

I used RTX4090. After loading the model and start to infer, it just output segmentation fault (core dumped), no other valuable info.

Image

@lingtengqiu
Copy link
Collaborator

lingtengqiu commented Apr 3, 2025

we update the codes, now we do not provide a config? do you use the lastest codes?

@liphao99
Copy link
Collaborator

liphao99 commented Apr 3, 2025

Before you updated the code, does it work properly? I just checked the latest version on cu12.1 and cu11.8 envs and it works fine. I think the exception isn't caused by repo update, because the code of pose_estimation is changed little.

@senbinyu
Copy link
Author

senbinyu commented Apr 7, 2025

I found that it can work on RTX A6000 device with the same code and env. It seems that the problem is caused by the device (4090 seems can not work properly). GPU memory of 4090 is abundant for LHM-500M, I am not sure what happened there.

@liphao99
Copy link
Collaborator

liphao99 commented Apr 8, 2025

This exception does not seem to be caused by our repo, since we have tested the code and environment on 4090. Could you repeated the error on another 4090?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

3 participants