Reproduce “Hand-Object Contact Transition Detection in 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

Huy Anh Nguyen, Chang Dong, Feras Dayoub, Minh Hoai

OASIS detects touch and release boundaries in first-person video by combining hand-focused and predicted object-region features with class-specific timing refinement. On ContactTransition, it scores 49.30 mAP without NMS, above the strongest baseline’s 43.93. It leads with and without suppression; ablations show lower scores without the object branch or with shared refinement. These benchmark results support component contributions, but do not establish object evidence as the sole cause.

Summary by alphaXiv.

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