May, 2024
3D 神经边缘重建
3D Neural Edge Reconstruction
Lei Li, Songyou Peng, Zehao Yu, Shaohui Liu, Rémi Pautrat...
TL;DRReal-world objects and environments predominantly consist of edge features. In this paper, the authors propose EMAP, a method for learning 3D edge representations using both lines and curves, utilizing Unsigned Distance Functions (UDF) from multi-view edge maps. The method achieves better 3D edge reconstruction and enhances neural surface reconstruction by capturing more details.