Revisiting Point Cloud Shape Classification with a Simple and Effective Baseline

TL;DR

This paper reveals that evaluation factors significantly influence point cloud classification performance and introduces SimpleView, a simple projection-based method that rivals or surpasses complex models on standard benchmarks.

Contribution

The paper uncovers the impact of evaluation factors on performance and proposes SimpleView, a straightforward approach that achieves competitive results with less complexity.

Findings

Evaluation schemes and data augmentation greatly affect performance metrics.

PointNet++ performs competitively when evaluation factors are controlled.

SimpleView outperforms state-of-the-art methods on key benchmarks.

Abstract

Processing point cloud data is an important component of many real-world systems. As such, a wide variety of point-based approaches have been proposed, reporting steady benchmark improvements over time. We study the key ingredients of this progress and uncover two critical results. First, we find that auxiliary factors like different evaluation schemes, data augmentation strategies, and loss functions, which are independent of the model architecture, make a large difference in performance. The differences are large enough that they obscure the effect of architecture. When these factors are controlled for, PointNet++, a relatively older network, performs competitively with recent methods. Second, a very simple projection-based method, which we refer to as SimpleView, performs surprisingly well. It achieves on par or better results than sophisticated state-of-the-art methods on ModelNet40 while being half the size of PointNet++. It also outperforms state-of-the-art methods on ScanObjectNN, a real-world point cloud benchmark, and demonstrates better cross-dataset generalization. Code is available at https://github.com/princeton-vl/SimpleView.