NVlabs / GLAMR

[CVPR 2022 Oral] Official PyTorch Implementation of "GLAMR: Global Occlusion-Aware Human Mesh Recovery with Dynamic Cameras”.
https://nvlabs.github.io/GLAMR
Other
359 stars 33 forks source link

Missing a 4th representation HybrIK(World) #35

Open AndreOliveira00 opened 1 year ago

AndreOliveira00 commented 1 year ago

Hi! Thanks for the excellent work.

In this gif: https://github.com/NVlabs/GLAMR/blob/main/docs/basketball_glamr_vs_hybrik.gif a direct comparison between GLAMR and HybrIK in world and camera projection was presented.

I analyzed global_recon/run_demo.py with some attention and didn't find a direct way to generate this 4th representation (HybrIK(World)), only GLAMR (Cam), GLAMR (World) and HybrIK (Cam).

My question is: Is there any way to get this 4th representation from the data collected in run_demo.py? Or should this process be performed while estimating the pose and shape generated by HybrIK in run_pose_est_on_video(...)?

I believe it would be interesting to include in the project the code snippet used to generate this 4th representation. And as such, I would be grateful if someone could publish it.

Thanks!