Variational relational point completion network for robust 3D classification

Variational relational point completion network for robust 3D classification

Real-scanned point clouds are often incomplete due to viewpoint, occlusion, and noise, which hampers 3D geometric modeling and perception. Existing point cloud completion methods tend to generate global shape skeletons and hence lack fine local details. Furthermore, they mostly learn a deterministic...

Full description

Bibliographic Details
Main Authors: Pan, Liang, Chen, Xinyi, Cai, Zhongang, Zhang, Junzhe, Zhao, Haiyu, Yi, Shuai, Liu, Ziwei
Other Authors: School of Computer Science and Engineering
Format: Journal Article
Language:English
Published: 2023
Subjects:
Engineering::Computer science and engineering
3D Perception
Point Cloud Completion
Online Access:https://hdl.handle.net/10356/172185

Internet

https://hdl.handle.net/10356/172185

Similar Items