English

Online Segmentation of LiDAR Sequences: Dataset and Algorithm

Computer Vision and Pattern Recognition 2022-07-22 v2

Abstract

Roof-mounted spinning LiDAR sensors are widely used by autonomous vehicles. However, most semantic datasets and algorithms used for LiDAR sequence segmentation operate on 360360^\circ frames, causing an acquisition latency incompatible with real-time applications. To address this issue, we first introduce HelixNet, a 1010 billion point dataset with fine-grained labels, timestamps, and sensor rotation information necessary to accurately assess the real-time readiness of segmentation algorithms. Second, we propose Helix4D, a compact and efficient spatio-temporal transformer architecture specifically designed for rotating LiDAR sequences. Helix4D operates on acquisition slices corresponding to a fraction of a full sensor rotation, significantly reducing the total latency. Helix4D reaches accuracy on par with the best segmentation algorithms on HelixNet and SemanticKITTI with a reduction of over 5×5\times in terms of latency and 50×50\times in model size. The code and data are available at: https://romainloiseau.fr/helixnet

Keywords

Cite

@article{arxiv.2206.08194,
  title  = {Online Segmentation of LiDAR Sequences: Dataset and Algorithm},
  author = {Romain Loiseau and Mathieu Aubry and Loïc Landrieu},
  journal= {arXiv preprint arXiv:2206.08194},
  year   = {2022}
}

Comments

Code and data are available at: https://romainloiseau.fr/helixnet

R2 v1 2026-06-24T11:53:53.749Z