Reproduce “Ego2Sim: Photorealistic Contact-Enabled Simulation from Egocentric Data” 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
Aidan Gildea, Sunwoong Choi
Ego2Sim turns a hands-free walk recorded with Meta Aria Gen 2 glasses into a simulator scene with rendered appearance and contact geometry. It builds a Gaussian splat from RGB video and a collider from the glasses’ SLAM points, aligned in one metric, gravity-aligned frame. Across five walks, collider F-scores reach 0.84–0.94, but MVS scores higher and takes longer to build.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.