Shape Reconstruction by Learning Differentiable Surface Representations
Résumé
Figure 1. Our method predicts a multi-patch representation where the patches are guaranteed not to collapse and to minimally overlap. Thanks to its explicit access to the surface's differential properties, our method computes normals and curvature analytically for any predicted point of the modeled surface.
Domaines
Informatique [cs]
Fichier principal
Bednarik_Shape_Reconstruction_by_Learning_Differentiable_Surface_Representations_CVPR_2020_paper.pdf (1.76 Mo)
Télécharger le fichier
| Origine | Fichiers éditeurs autorisés sur une archive ouverte |
|---|