Reproduce “Open-Vocabulary Functional 3D Human-Scene Interaction Generation” 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
Jie Liu, Yu Sun, Alpar Cseke, Yao Feng, Nicolas Heron, Michael J. Black, Yan Zhang
FunHSI infers the functional scene element and body contacts implied by a task prompt and synthesizes a 3D human pose. It uses vision-language and language models to build a contact graph, inpaints an image for pose initialization, and refines the pose in the scene. On 60 curated tasks, functional contact distance is 0.2968 versus 0.4199 for GenHSI*, on this benchmark.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.