Reproduce “DreamHMR: From Video Generation to Human Motion Modeling via Structural Focusing” 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
Zhengjie Xu, Ruihao Gong, Xianglong Liu
DreamHMR adapts a video model to predict complete human renderings and recover bodies and hands from video using whole-body context and evidence-gated local features. On held-out occlusion sequences, Wan-5B records 107.1 mm body and 55.1 mm hand MPJPE, versus 115.2 mm and 61.2 mm for 3DB-DINOv3. This supports reconstruction under tested protocols, not guaranteed recovery of every hidden pose or hand.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.