Universal category-agnostic 9D object pose estimation
Track and visualize points in a video
ICLR 2025 demo of multiview-equivariant 3D correspondence