Research map / Point clouds / Segmentation and detection
PUPS: Point Cloud Unified Panoptic Segmentation
Shihao Su, Jianyun Xu, Huanyu Wang, Zhenwei Miao, Xin Zhan, Dayang Hao, Xi Li
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 |
|---|---|---|---|
| Add classifier refinement (ablation) | PUPS M1 · SemanticKITTI validation | PQ: 44.5 -> 50.9 | Improved |
| Add context-aware CutMix (ablation) | PUPS M1 · SemanticKITTI validation | PQ: 44.5 -> 55.9 | Improved |
| Use classifier refinement and context-aware CutMix together (ablation) | PUPS · SemanticKITTI validation | PQ: 55.9 -> 64.4 | Improved |
| Use 3 rather than 1 refinement stage (ablation) | PUPS · SemanticKITTI validation | PQ: 62.1 -> 64.4 | Improved |
| Increase refinement stages from 3 to 4 (ablation) | PUPS · SemanticKITTI validation | PQ: 64.4 -> 63.2 | Hurt |
| Increase refinement stages from 3 to 5 (ablation) | PUPS · SemanticKITTI validation | PQ: 64.4 -> 63.0 | Hurt |
| Use 100 rather than 50 classifiers (ablation) | PUPS · SemanticKITTI validation | PQ: 63.3 -> 64.4 | Improved |
| Increase the classifier count from 100 to 150 (ablation) | PUPS · SemanticKITTI validation | PQ: 64.4 -> 63.7 | Hurt |
| Increase the classifier count from 100 to 200 (ablation) | PUPS · SemanticKITTI validation | PQ: 64.4 -> 63.8 | Hurt |
| Replace random CutMix with context-aware CutMix (ablation) | PUPS M1 · SemanticKITTI validation | PQ: 50.8 -> 55.9 | Improved |
| Use PUPS instead of Panoptic-PHNet | SemanticKITTI test · SemanticKITTI test | PQ: 61.5 -> 62.2 | Improved |
| Use PUPS instead of GP-S3Net | SemanticKITTI validation · SemanticKITTI validation | PQ: 63.3 -> 64.4 | Improved |
Later papers that compared with this method
1 later accepted paper reports PUPS as a baseline in their results tables; 0 of them report a better number on most of the results they share with it.
- LiDAR-Camera Panoptic Segmentation via Geometry-Consistent and Semantic-Aware Alignment (ICCV 2023 · better on 1 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Panoptic-PolarNet: Proposal-Free LiDAR Point Cloud Panoptic Segmentation (CVPR 2021 · similarity 0.90)
- OneFormer3D: One Transformer for Unified Point Cloud Segmentation (CVPR 2024 · similarity 0.88)
- Panoptic-PHNet: Towards Real-Time and High-Precision LiDAR Panoptic Segmentation via Clustering Pseudo Heatmap (CVPR 2022 · similarity 0.88)
- Lidar Panoptic Segmentation in an Open World (IJCV 2024 · similarity 0.88)
- FreePoint: Unsupervised Point Cloud Instance Segmentation (CVPR 2024 · similarity 0.87)
- Beyond the Label Itself: Latent Labels Enhance Semi-supervised Point Cloud Panoptic Segmentation (AAAI 2024 · 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.
