Reproduce “PSGAgent: Enhancing MLLMs' Dynamic Visual Perception for Panoptic Scene Graph Generation via Agentic 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
Qingyun Liu, Yuankun Liu, Jiaye Feng, Jie Huang, Zonghong Dai, Haochen Li, Jingkun Wang, Weiping Li and 1 more
PSGAgent generates panoptic scene graphs by letting a vision-language model zoom into regions or highlight object pairs when relationships are uncertain. It is trained with error-aware examples and reinforcement learning to choose these visual tools. On PSG SGDet, its 3B model scores 46.58% Mean versus 42.98% for 3B SGG-R3; this benchmark result does not verify every predicted relation.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.