TerraSeg: Self-Supervised Ground Segmentation for Any LiDAR
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2026
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866910082098266112 |
|---|---|
| author | Lentsch, Ted Montiel-Marín, Santiago Caesar, Holger Gavrila, Dariu M. |
| author_facet | Lentsch, Ted Montiel-Marín, Santiago Caesar, Holger Gavrila, Dariu M. |
| contents | LiDAR perception is fundamental to robotics, enabling machines to understand their environment in 3D. A crucial task for LiDAR-based scene understanding and navigation is ground segmentation. However, existing methods are either handcrafted for specific sensor configurations or rely on costly per-point manual labels, severely limiting their generalization and scalability. To overcome this, we introduce TerraSeg, the first self-supervised, domain-agnostic model for LiDAR ground segmentation. We train TerraSeg on OmniLiDAR, a unified large-scale dataset that aggregates and standardizes data from 12 major public benchmarks. Spanning almost 22 million raw scans across 15 distinct sensor models, OmniLiDAR provides unprecedented diversity for learning a highly generalizable ground model. To supervise training without human annotations, we propose PseudoLabeler, a novel module that generates high-quality ground and non-ground labels through self-supervised per-scan runtime optimization. Extensive evaluations demonstrate that, despite using no manual labels, TerraSeg achieves state-of-the-art results on nuScenes, SemanticKITTI, and Waymo Perception while delivering real-time performance. Our code and model weights are publicly available. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2603_27344 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | TerraSeg: Self-Supervised Ground Segmentation for Any LiDAR Lentsch, Ted Montiel-Marín, Santiago Caesar, Holger Gavrila, Dariu M. Computer Vision and Pattern Recognition 68T10, 62H35, 68T05, 68U10 I.2.10; I.4.8; I.5.1; I.5.4 LiDAR perception is fundamental to robotics, enabling machines to understand their environment in 3D. A crucial task for LiDAR-based scene understanding and navigation is ground segmentation. However, existing methods are either handcrafted for specific sensor configurations or rely on costly per-point manual labels, severely limiting their generalization and scalability. To overcome this, we introduce TerraSeg, the first self-supervised, domain-agnostic model for LiDAR ground segmentation. We train TerraSeg on OmniLiDAR, a unified large-scale dataset that aggregates and standardizes data from 12 major public benchmarks. Spanning almost 22 million raw scans across 15 distinct sensor models, OmniLiDAR provides unprecedented diversity for learning a highly generalizable ground model. To supervise training without human annotations, we propose PseudoLabeler, a novel module that generates high-quality ground and non-ground labels through self-supervised per-scan runtime optimization. Extensive evaluations demonstrate that, despite using no manual labels, TerraSeg achieves state-of-the-art results on nuScenes, SemanticKITTI, and Waymo Perception while delivering real-time performance. Our code and model weights are publicly available. |
| title | TerraSeg: Self-Supervised Ground Segmentation for Any LiDAR |
| topic | Computer Vision and Pattern Recognition 68T10, 62H35, 68T05, 68U10 I.2.10; I.4.8; I.5.1; I.5.4 |
| url | https://arxiv.org/abs/2603.27344 |