Research map / Detection and segmentation / CNN features
Lidar Panoptic Segmentation in an Open World
Anirudh Srinivasan Chakravarthy, Meghana Reddy Ganesina, Peiyun Hu, Laura Leal-Taixé, Shu Kong, Deva Ramanan, Aljoša Ošep
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use class-agnostic hierarchical grouping instead of class-specific (ablation) | OWL · SemanticKITTI val (closed world) | Prec Th: 79.7 -> 79.9 | Improved |
| add holistic second-stage objectness classifier (ablation) | OWL · SemanticKITTI val (closed world) | PQ: 58.7 -> 58.8 | Improved |
| add regression loss for objectness (ablation) | OWL · SemanticKITTI val (closed world) | PQ: 58.8 -> 58.9 | Improved |
| add semantic majority vote within each segment (ablation) | OWL · SemanticKITTI val (closed world) | PQ: 58.9 -> 59.0 | Improved |
| replace learned semantic classifier with GT semantic map | OWL · SemanticKITTI val (closed world) | PQ: 59.0 -> 98.3 | Improved |
| use perfect objectness oracle scores | OWL · SemanticKITTI val (closed world) | PQ: 59.0 -> 59.2 | Improved |
| use OWL grouping over 4D-PLS inference | 4D-PLS · SemanticKITTI val (closed world) | PQ: 56.5 -> 58.7 | Improved |
| use OWL instead of 4D-PLS for known classes | 4D-PLS · KITTI360 (cross-dataset), Vocabulary 1 known | PQ: 56.1 -> 59.4 | Improved |
| use OWL instead of 4D-PLS for unknown instances | 4D-PLS · KITTI360 (cross-dataset), Vocabulary 1 unknown | Recall: 2.0 -> 45.1 | Improved |
| lower center-objectness threshold for other-class points at inference (ablation) | 4D-PLS · KITTI360 (cross-dataset), Vocabulary 1 unknown | Recall: 2.0 -> 6.0 | Improved |
| use separate learned grouping for thing classes and hierarchical segmentation for other (ablation) | OWL · KITTI360 (cross-dataset), Vocabulary 1 unknown | UQ: 36.3 -> 36.6 | No clear effect |
| group rare classes into catch-all other (Vocabulary 1 vs 2) | OWL · KITTI360 (cross-dataset) | PQ: 53.4 -> 59.4 | Improved |
Closest papers by meaning
- Panoptic-PolarNet: Proposal-Free LiDAR Point Cloud Panoptic Segmentation (CVPR 2021 · similarity 0.91)
- LiDAR-Based Panoptic Segmentation via Dynamic Shifting Network (CVPR 2021 · similarity 0.91)
- Panoptic-PHNet: Towards Real-Time and High-Precision LiDAR Panoptic Segmentation via Clustering Pseudo Heatmap (CVPR 2022 · similarity 0.91)
- Open-World Semantic Segmentation for LIDAR Point Clouds (ECCV 2022 · similarity 0.90)
- 3D-AVS: LiDAR-based 3D Auto-Vocabulary Segmentation (CVPR 2025 · similarity 0.88)
- No Thing, Nothing: Highlighting Safety-Critical Classes for Robust LiDAR Semantic Segmentation in Adverse Weather (CVPR 2025 · similarity 0.88)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
