Reproduce “PIVOT: Physics-Grounded Verification for AI-Generated Audio-Video Detection” 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
Bo Zheng, Kangran Zhao, Xiaoyu Zhang, Weinan Guan, Zhiheng Li, Yize Chen, Haizhou Li, Qingshan Liu and 2 more
PIVOT checks whether events in audio-video clips follow relevant physical constraints, using captions, estimated visual and audio quantities, and selected law checks to return labels with evidence. On a test split, it reached 70.30% accuracy and 64.29% F1 on Real+Seedance, versus Gemini 3.1 Pro’s 53.96% and 60.09%. Evidence is limited to nine event families and two generators.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.