Reproduce “Zero-Shot Object Counting with Language-Vision Models” 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

Jingyi Xu, Hieu Le, Dimitris Samaras

Patch-Sel lets users name a counting target and selects image crops as exemplars for a class-agnostic counter. It uses Stable Diffusion-generated patches to form an image-specific prototype, retrieve relevant candidates, and an error predictor to rank them. On FSC-147’s test split, it reports mean absolute error of 21.53, versus 29.25 for top-three objectness-ranked RPN proposals; this result does not establish deployment reliability.

Summary by alphaXiv.

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