PRIMA animal 3D mesh recovery accepted to WACV2027 with 9K pseudo-3D dataset
TrackingActions · x · 2026-11-03
PRIMA (PRIors for Mesh Adaptation), from Mackenzie Mathis's lab, has been accepted to WACV2027; the updated preprint and SOTA single-image-to-3D animal shape & pose models are now available for testing.
The paper tackles monocular 3D animal mesh recovery, where long-tailed species distribution and scarce 3D supervision cause poor generalization:
- Taxonomic shape priors derived from BioCLIP features inject species-level knowledge into mesh reconstruction;
- Adaptation-driven data bootstrapping converts abundant 2D animal annotations into reliable pseudo-3D supervision via lightweight instance-level refinement and quality-aware filtering, yielding Quadruped3D, a dataset of 9K pseudo-3D annotations from real-world images.
The framework achieves state-of-the-art results on Animal3D, CtrlAni3D, Quadruped2D, and Animal Kingdom, with applications in biomedical research and avatars.
More from Embodied
- ETH Zurich's ADAPT runs diffusion action priors for real-time language-driven humanoid control at 50Hz — breadli428 · 2026-11-03
- ADAPT architecture breakdown: diffusion Transformer plus residual RL in four stages — breadli428 · 2026-11-03
- Xiaomi open-sources Robotics-1, a robot foundation model trained on 100K hours of trajectories — CyberRobooo · 2026-11-03
- Factory photos show desktop robot modules in volume assembly, project hits production stage — victormustar · 2026-11-03
- Machenta demos instant robot learning from passersby at London King's Cross — Ed__Johns · 2026-11-03
- Engineer Ports NVIDIA's Open-Source SONIC Model to AGIBOT X2 in 6 GPU-Hours — Scobleizer · 2026-11-03