Design equivariant neural networks for 3D point cloud

TL;DR

This paper introduces a simple, low-complexity method to incorporate group equivariance into state-of-the-art 3D point cloud neural networks, enhancing their performance and robustness while reducing implementation difficulty.

Contribution

A general procedure for adding group equivariance to 3D point cloud models that balances performance, complexity, and ease of implementation.

Findings

Outperforms existing group equivariant models in accuracy and complexity.

Improves mIoU in semantic segmentation tasks.

Achieves better results with finite-rotation equivariance and augmentation.

Abstract

This work seeks to improve the generalization and robustness of existing neural networks for 3D point clouds by inducing group equivariance under general group transformations. The main challenge when designing equivariant models for point clouds is how to trade-off the performance of the model and the complexity. Existing equivariant models are either too complicate to implement or very high complexity. The main aim of this study is to build a general procedure to introduce group equivariant property to SOTA models for 3D point clouds. The group equivariant models built form our procedure are simple to implement, less complexity in comparison with the existing ones, and they preserve the strengths of the original SOTA backbone. From the results of the experiments on object classification, it is shown that our methods are superior to other group equivariant models in performance and complexity. Moreover, our method also helps to improve the mIoU of semantic segmentation models. Overall, by using a combination of only-finite-rotation equivariance and augmentation, our models can outperform existing full $SO(3)$-equivariance models with much cheaper complexity and GPU memory. The proposed procedure is general and forms a fundamental approach to group equivariant neural networks. We believe that it can be easily adapted to other SOTA models in the future.