Reproduce “Promptable Animal Pose Tracking Across Species” 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

Le Li, Daniela Ivanova, Nicolas Pugeault

The authors present a two-route method for tracking researcher-selected animal keypoints from one annotated video frame. One route trains a prompt-conditioned matcher, while the other matches pretrained visual features without task-specific training; both still require the reference-frame annotation. On APTv2, supervised results cross HRNet-W48’s scores across thresholds, and held-out-family results show weaker transfer for supervised tracking.

Summary by alphaXiv.

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