Reproduce “EgoReveal: Learning Active Vision and Object Manipulation under Partial Occlusions from Egocentric Demonstrations” 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
Hiromu Kurihara, Kazutoshi Tanaka, Masashi Hatano, Tadashi Kozuno, Hideo Saito, Mariko Isogawa
EgoReveal learns camera repositioning and object manipulation from human smart-glasses demonstrations, switching to manipulation when visibility is resolved. In the partially occluded Soft Toy task, it achieved 85% success, versus 35% for EgoZero without active vision and 55% without visibility-based switching. Success means stable grasp and lift; the results concern targets partly visible at the start.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.