What Can be Seen is What You Get: Structure Aware Point Cloud Augmentation
This work addresses the need for efficient data augmentation in point cloud segmentation, particularly for lidar-based applications, though it is incremental as it builds on existing augmentation techniques.
The paper tackles the problem of limited labeled data for point cloud semantic segmentation by introducing sensor-centric augmentation methods that diversify datasets while preserving data structure, resulting in improved performance across multiple neural networks on the SemanticKITTI dataset and enabling the use of very small datasets to save time and costs.
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.