Reproduce “Exploiting Textual Semantics for Robust Cross-View Object Correspondence” 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
Bing Fan, Yunhe Feng, Yan Huang, Heng Fan
TeSCo uses descriptions of a marked object to guide its search for that object in video from another camera. It adapts descriptions to search-view features and fuses them with visual features for segmentation. On Ego-Exo4D, it reports 65.35% mean intersection-over-union, versus 55.50% for a baseline without text modules. Occlusion, similar distractors, and small targets remain failure cases.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.