Reproduce “EgoTac: In-the-wild Tactile Prediction from Egocentric Vision” 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

Wenkang Zhang, Chengbo Yuan, Zicheng Zhang, Zhengxue Cheng, Yang Gao

EgoTac predicts per-vertex contact and force from egocentric video, using measured force data alongside contact-only labels from hand–object datasets. A hand-surface representation lets examples with different labels train the predictor. On held-out EgoTac-SC episodes, active-region force MAE is 0.056 N; on OAKINK2, contact F1 is 0.435 versus HACO’s 0.400. OpenTouch evaluates normalized relative dynamics, while unlabeled video results remain qualitative.

Summary by alphaXiv.

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