Any proper guide for training human pose in Nvidia Isaac gym Thanks!

I want to train pose in isaac gym using
https://catalog.ngc.nvidia.com/orgs/nvidia/teams/tao/models/bodyposenet
Using rainforcement methode, if there any related work please suggest me thanks you.

Hi @pantakan.totae,

I’m afraid that RL isn’t really suitable for body pose estimation. A more closely related work for RL might be AMP (Adversarial Motion Priors - [2104.02180] AMP: Adversarial Motion Priors for Stylized Physics-Based Character Control).

We have AMP as an example in IsaacGymEnvs:

Take care,
-Gav

1 Like

Sorry for misunderstood you because of my poor English i mean extract motion data from pose estimation and using it in training Agent

If you want to use human pose estimation derived motion data, you should definitely review the AMP example. It uses sample data from mocap libraries and the agent learns to mimic the motion.

Take care,
-Gav

1 Like

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.