SEMAGIC: Learning Semantically Consistent Deformable 3D Representations from In-the-Wild Images 事件

SEMAGIC: Learning Semantically Consistent Deformable 3D Representations from In-the-Wild Images arXiv:2605.27938v1 Announce Type: new Abstract: Learning deformable 3D object models from single-view in-the-wild images has enabled impressive 3D shape reconstruction without supervision. However, it remains unclear whether these models capture the semantic structure required for downstream tasks. We find that existing deformable reconstruction approaches, despite producing visually plausible geomet