Reproduce “FaVChat: Hierarchical Prompt-Query Guided Facial Video Understanding with Data-Efficient GRPO” 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
Fufangchen Zhao, Songbai Tan, Xuerui Qiu, Linrui Xun, Wenhao Jiang, Jinkai Zheng, Hehe Fan, Jian Gao and 2 more
FaVChat lets questions guide visual evidence across general and facial video representations, using staged feature selection and text-conditioned weighting. Its DE-GRPO training method prioritizes videos according to their learning utility. On DFEW, it reports 70.01 UAR and 79.79 WAR in zero-shot visual-only evaluation; this does not establish that facial expressions reveal internal emotional states.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.