Reproduce “EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience” 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

Xinhao Ji, Hui En Pang, Hao Li, Kairui Hu, Jiner Wang, Mingze Sun, Ziang Cao, Yalun Dai and 7 more

EgoX jointly reconstructs camera motion, scene depth, hands, and rigid objects from monocular video in a shared metric frame. It trains these predictions together using aligned annotations and predicted cameras to place hand and object geometry in the world frame. On held-out EgoX-Bench, it reports 113.24 mm W-MPJPE versus 328.81 mm for WiLoR + SLAM, and assumes rigid objects.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.