Reproduce “Detect in Any Scene: An Agentic Framework for Object Detection with Experience-Aware Reasoning” 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
Wenlun Zhang, Jun Yin, Kentaro Yoshioka
DetAS uses a vision-language agent to choose image restoration and specialist detectors for each scene, while DetAS-X adds experience retrieved from 50 annotated training examples per dataset. Across six benchmarks, DetAS-X reports a mean F1 of 47.99%, versus 19.63% for the strongest MLLM baseline. This arithmetic mean reflects tested datasets, not universal transfer to unseen domains.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.