Research map / Point clouds / Segmentation and detection
4D Panoptic LiDAR Segmentation
Check your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Process multiple scans jointly in a 4D volume instead of one scan (ablation) | 4D-PLS · SemanticKITTI validation | 4D-PLS: 51.92 -> 62.74 | Helped |
| Sample past points proportional to objectness rather than propagating only thing points (ablation) | 4D-PLS · SemanticKITTI validation, 4 scans | 4D-PLS: 60.94 -> 62.74 | Helped |
| Use temporal decay instead of importance sampling (ablation) | 4D-PLS · SemanticKITTI validation, 4 scans | 4D-PLS: 62.74 -> 62.14 | Hurt |
| Use temporal stride instead of importance sampling (ablation) | 4D-PLS · SemanticKITTI validation, 4 scans | 4D-PLS: 62.74 -> 61.63 | Hurt |
| Increase the temporal window from 2 to 4 scans (ablation) | 4D-PLS · SemanticKITTI validation | 4D-PLS: 59.86 -> 62.74 | Helped |
| Increase the temporal window from 4 to 8 scans (ablation) | 4D-PLS · SemanticKITTI validation | 4D-PLS: 62.74 -> 59.09 | Hurt |
| Combine learned embeddings with spatio-temporal coordinates rather than use embeddings alone (ablation) | 4D-PLS · SemanticKITTI validation, 2 scans | 4D-PLS: 48.58 -> 59.86 | Helped |
| Combine learned embeddings with spatio-temporal coordinates rather than use embeddings alone (ablation) | 4D-PLS · SemanticKITTI validation, 4 scans | 4D-PLS: 56.77 -> 62.74 | Helped |
| Combine learned embeddings with spatial coordinates rather than use embeddings alone (ablation) | 4D-PLS · SemanticKITTI validation, 2 scans | 4D-PLS: 48.58 -> 54.15 | Helped |
| Combine learned embeddings with spatial coordinates rather than use embeddings alone (ablation) | 4D-PLS · SemanticKITTI validation, 4 scans | 4D-PLS: 56.77 -> 58.43 | Helped |
| Use 4D-PLS jointly rather than single-scan prediction with tracking-by-detection | SemanticKITTI test · SemanticKITTI test | 4D-PLS: 40.18 -> 56.89 | Helped |
| Increase the temporal window from 2 to 4 scans (ablation) | 4D-PLS · SemanticKITTI validation | S_cls: 60.95 -> 60.46 | Hurt |
Closest papers by meaning
- Zero-Shot 4D Lidar Panoptic Segmentation (CVPR 2025 · similarity 0.90)
- Is Your LiDAR Placement Optimized for 3D Scene Understanding? (NeurIPS 2024 · similarity 0.88)
- HiLoTs: High-Low Temporal Sensitive Representation Learning for Semi-Supervised LiDAR Segmentation in Autonomous Driving (CVPR 2025 · similarity 0.88)
- Towards Learning to Complete Anything in Lidar (ICML 2025 · similarity 0.88)
- 3D-AVS: LiDAR-based 3D Auto-Vocabulary Segmentation (CVPR 2025 · similarity 0.87)
- TREND: Unsupervised 3D Representation Learning via Temporal Forecasting for LiDAR Perception (NeurIPS 2025 · similarity 0.87)
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.