AI RESEARCH
AnyLift: Scaling Motion Reconstruction from Internet Videos via 2D Diffusion
arXiv CS.CV
•
ArXi:2604.17818v1 Announce Type: new Reconstructing 3D human motion and human-object interactions (HOI) from Internet videos is a fundamental step toward building large-scale datasets of human behavior. Existing methods struggle to recover globally consistent 3D motion under dynamic cameras, especially for motion types underrepresented in current motion-capture datasets, and face additional difficulty recovering coherent human-object interactions in 3D. We