
A unified diffusion framework, S2Fusion, that combines scene geometry and sparse tracking signals to generate plausible and coherent full-body human motions, overcoming the inherent ambiguities in the sparse-to-dense mapping problem.


coremsg

estimating-full-body-human-motion-in-3d-scenes-from-sparse-tracking-signals-using-a-unified-diffusion-framework


Estimating Full-Body Human Motion in 3D Scenes from Sparse Tracking Signals using a Unified Diffusion Framework


title_rewrite


PhysPT, a Transformer encoder-decoder model, improves the physical plausibility of kinematics-based 3D human motion estimates and infers motion forces from monocular videos without requiring 3D annotated training data.


estimating-human-dynamics-from-monocular-videos-with-physics-aware-pretrained-transformer


Estimating Human Dynamics from Monocular Videos with Physics-aware Pretrained Transformer
