Reproduce “EgoHSI3R: Online 4D Hand-Scene Interaction Reconstruction from Egocentric Video” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Mingzhi Pei, Zhangchen Ye, Ruoqu Chen, Liu Cao, Mengdi Xu
EgoHSI3R reconstructs hands, scene, camera motion, and predicted contact together in a shared world as video frames arrive. Using prior frames and the current image, it updates a recurrent scene representation and refines hand estimates with local scene cues and contact feedback. On ARCTIC, it achieves 108.80 mm W-MPJPE versus 124.59 mm for StableHand, but has worse acceleration and relative hand-position errors.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.