PointCVaR: Risk-optimized Outlier Removal for 3D Robust Point Cloud Classification

Xinke Li, Junchi Lu, Henghui Ding, Changsheng Sun, Joey Tianyi Zhou, Yeow Meng Chee

Paper

Code

Appendix

Overview

Figure. Visualization of Point Risks for Noisy Point Clouds

What is point risk?

Conditional Value-at-risk (CVaR)

Figure. Illustrative diagram of Point Risk Distribution and CVaR

CVaR Optimization

Figure. The proposed framework of outlier removal by PointCVaR. Point risks are obtained by entering the noise sample into a trained classification model. Subsequently, an optimization problem is solved to minimize the tail of risk distribution, which leads to binary weights for noise point removal. The processed point cloud is utilized for classification.