What Can be Seen is What You Get: Structure Aware Point Cloud Augmentation

TL;DR

This paper introduces structure-aware point cloud augmentation techniques that enhance dataset diversity for semantic segmentation, improving neural network performance and enabling effective training on smaller datasets.

Contribution

The paper proposes novel sensor-centric augmentation methods that preserve lidar data structure, allowing for dataset enrichment and scene creation, which improves model training efficiency.

Findings

All tested neural networks showed performance improvements.

Augmentation enables effective training on smaller datasets.

Methods save annotation and training costs.

Abstract

To train a well performing neural network for semantic segmentation, it is crucial to have a large dataset with available ground truth for the network to generalize on unseen data. In this paper we present novel point cloud augmentation methods to artificially diversify a dataset. Our sensor-centric methods keep the data structure consistent with the lidar sensor capabilities. Due to these new methods, we are able to enrich low-value data with high-value instances, as well as create entirely new scenes. We validate our methods on multiple neural networks with the public SemanticKITTI dataset and demonstrate that all networks improve compared to their respective baseline. In addition, we show that our methods enable the use of very small datasets, saving annotation time, training time and the associated costs.