I run infer_wild_mesh.py and use J_regressor in SMPL_NEUTRAL.pkl to get 24 joints psitions, the performance is great, but the root seems to be fixed at about (0, 0, 0). If I want to get joint positions in global corridnate, should I run infer_wild.py and select the root position as global transl?
Hi, thanks for the amazing work!
I run
infer_wild_mesh.py
and use J_regressor in SMPL_NEUTRAL.pkl to get 24 joints psitions, the performance is great, but the root seems to be fixed at about (0, 0, 0). If I want to get joint positions in global corridnate, should I runinfer_wild.py
and select the root position as global transl?