Research map / Detection and segmentation / 3D detection and driving
Panoptic-PHNet: Towards Real-Time and High-Precision LiDAR Panoptic Segmentation via Clustering Pseudo Heatmap
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 |
|---|---|---|---|
| Use Panoptic-PHNet instead of EfficientLPS | LiDAR panoptic segmentation · SemanticKITTI test | PQ: 57.4 -> 61.5 | Helped |
| Use Panoptic-PHNet instead of (AF)²-S3Net + CenterPoint | LiDAR panoptic segmentation · nuScenes test | PQ: 76.8 -> 80.1 | Helped |
| Use Panoptic-PHNet instead of Panoptic-PolarNet | LiDAR panoptic segmentation · nuScenes validation | PQ: 63.4 -> 74.7 | Helped |
| Add the center grouping module (ablation) | Panoptic-PHNet clustering pseudo heatmap · SemanticKITTI validation | PQ^Th: +4.4% PQ^Th | Helped |
| Add the knn-transformer (ablation) | Panoptic-PHNet instance branch · SemanticKITTI validation | PQ^Th: +1.2% PQ^Th | Helped |
| Replace learned heatmap with clustering pseudo heatmap (ablation) | Panoptic-PHNet · SemanticKITTI validation | PQ^Th: 67.9 -> 69.3 (+1.4) | Helped |
| Replace learned heatmap with clustering pseudo heatmap (ablation) | Panoptic-PHNet · nuScenes validation | PQ^Th: 65.7 -> 72.5 (+6.8) | Helped |
| Add fine-grained voxel features to the semantic branch (ablation) | Panoptic-PHNet · SemanticKITTI validation | mIoU: 64.5 -> 65.7 | Helped |
| Add fine-grained voxel features to the instance branch (ablation) | Panoptic-PHNet · SemanticKITTI validation | PQ^Th: 69.0 -> 69.3 | Helped |
| Use axis-aligned instance centers instead of mass centers (ablation) | Panoptic-PHNet · SemanticKITTI validation | PQ: 61.6 -> 61.7 | No clear effect |
| Use axis-aligned instance centers instead of mass centers (ablation) | Panoptic-PHNet · nuScenes validation | PQ: 72.6 -> 74.7 | Helped |
| Use annotated box centers instead of axis-aligned centers (ablation) | Panoptic-PHNet · nuScenes validation | PQ: 74.7 -> 74.8 | Helped |
Closest papers by meaning
- Panoptic-PolarNet: Proposal-Free LiDAR Point Cloud Panoptic Segmentation (CVPR 2021 · similarity 0.94)
- LiDAR-Based Panoptic Segmentation via Dynamic Shifting Network (CVPR 2021 · similarity 0.93)
- Sparse Cross-Scale Attention Network for Efficient LiDAR Panoptic Segmentation (AAAI 2022 · similarity 0.90)
- PC-BEV: An Efficient Polar-Cartesian BEV Fusion Framework for LiDAR Semantic Segmentation (AAAI 2025 · similarity 0.88)
- How Do Images Align and Complement LiDAR? Towards a Harmonized Multi-modal 3D Panoptic Segmentation (ICML 2025 · similarity 0.88)
- Beyond the Label Itself: Latent Labels Enhance Semi-supervised Point Cloud Panoptic Segmentation (AAAI 2024 · 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.