HaMeR
geopavlakos
Transformer-based reconstruction of three-dimensional human hand meshes from images.
What you can explore
Explore visual hand observations as an input to human-motion capture or retargeting research.
Before you use it
Image reconstruction is not robot control. Occlusion and scale ambiguity need evaluation, and MANO and checkpoint terms must be checked.
Original contributors
Georgios Pavlakos, Dandan Shan, Ilija Radosavovic, Angjoo Kanazawa, David Fouhey and Jitendra Malik
License / access: Code: MIT · data, models and dependencies may have separate terms
Editorial reference · checked 2026-09-22