InfiNoVA: Infinite Novel View Augmentation for Viewpoint Invariant Robot Policies
InfiNoVA is introduced, a data-augmentation framework that converts synchronized multi-camera demonstrations into a dense distribution of geometrically consistent training views that improves frame-level fidelity and temporal consistency while reducing task-critical hallucinations observed in generative novel-view synt...
Sai Puneeth Reddy Gottam, Elmar Rueckert, Vedant Dave
· 0 citations