Regularization Strategy for Point Cloud via Rigidly Mixed Sample

TL;DR

This paper introduces RSMix, a novel data augmentation method for point clouds that enhances regularization by mixing shape-preserved subsets from different samples, improving shape classification performance.

Contribution

The paper proposes RSMix, a new augmentation technique specifically designed for point clouds that maintains structural integrity while generating diverse training samples.

Findings

RSMix significantly improves shape classification accuracy.

Combining RSMix with other augmentations yields further performance gains.

RSMix effectively regularizes deep neural networks for point cloud data.

Abstract

Data augmentation is an effective regularization strategy to alleviate the overfitting, which is an inherent drawback of the deep neural networks. However, data augmentation is rarely considered for point cloud processing despite many studies proposing various augmentation methods for image data. Actually, regularization is essential for point clouds since lack of generality is more likely to occur in point cloud due to small datasets. This paper proposes a Rigid Subset Mix (RSMix), a novel data augmentation method for point clouds that generates a virtual mixed sample by replacing part of the sample with shape-preserved subsets from another sample. RSMix preserves structural information of the point cloud sample by extracting subsets from each sample without deformation using a neighboring function. The neighboring function was carefully designed considering unique properties of point cloud, unordered structure and non-grid. Experiments verified that RSMix successfully regularized the deep neural networks with remarkable improvement for shape classification. We also analyzed various combinations of data augmentations including RSMix with single and multi-view evaluations, based on abundant ablation studies.