SAT: Supervisor Regularization and Animation Augmentation for Two-process Monocular Texture 3D Human Reconstruction
ACM International Conference on Multimedia (ACM MM), 2025
Abstract
SAT addresses the difficulty of learning a textured 3D human from one front-view image. It combines several geometric priors in a two-process reconstruction framework. Feature regularization helps the model learn from multi-view supervision, while online animation augmentation expands the available training examples.
