Reproduce “OmniHands: Towards Robust 4D Hand Mesh Recovery via A Versatile Transformer” 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
Dixuan Lin, Yuxiang Zhang, Mengcheng Li, Wei Jing, Qi Yan, Qianying Wang, Yebin Liu, Hongwen Zhang
OmniHands recovers 3D hand meshes from images by cropping each hand separately and encoding its spatial relation to the other hand. It uses shared attention to combine hand features with video or multi-view context. On unseen ARCTIC, its single-image InterHand2.6M-only model reports 14.21 mm Ra-MPJPE after root alignment and skeleton scaling, versus 22.74 mm for InterWild, for cases where both hands are visible.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.