Reproduce “Towards Metric-Aware Multi-Person Mesh Recovery by Jointly Optimizing Human Crowd in Camera Space” 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

Kaiwen Wang, Kaili Zheng, Yiming Shi, Chenyi Guo, Ji Wu

The authors introduce DTO, a joint optimization method that refines people’s 3D placements in images using demographic height priors and monocular depth cues, then use its labels to build DTO-Humans and train a metric-aware human mesh recovery model. On Relative Human, DTO raises CameraHMR’s PCDR₀.₂ from 60.43 to 74.16 without benchmark-specific fine-tuning, but depends on depth-map quality.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.