Research map / Detection and segmentation / 3D detection and driving
CAGroup3D: Class-Aware Grouping for 3D Object Detection on Point Clouds
Read the paperCheck 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 CAGroup3D instead of FCAF3D | 3D object detection · ScanNet V2 | mAP@0.25: 71.5 -> 75.1 | Helped |
| Use CAGroup3D instead of FCAF3D | 3D object detection · ScanNet V2 | mAP@0.5: 57.3 -> 61.3 | Helped |
| Use CAGroup3D instead of FCAF3D | 3D object detection · SUN RGB-D | mAP@0.25: 64.2 -> 66.8 | Helped |
| Use CAGroup3D instead of FCAF3D | 3D object detection · SUN RGB-D | mAP@0.5: 48.9 -> 50.2 | Helped |
| Add voxel-wise semantic prediction (ablation) | fully sparse VoteNet baseline on ScanNet V2 val · ScanNet V2 val | mAP@0.25: 68.22 -> 69.24 | Helped |
| Add class-specific diverse local groups (ablation) | semantic-prediction grouping on ScanNet V2 val · ScanNet V2 val | mAP@0.25: 68.22 -> 72.10 | Helped |
| Add RoI-Conv proposal refinement (ablation) | CAGroup3D on ScanNet V2 val · ScanNet V2 val | mAP@0.5: 57.18 -> 60.31 | Helped |
| Replace FPN-based ResNet with BiResNet (ablation) | CAGroup3D on ScanNet V2 val · ScanNet V2 val | mAP@0.25: 72.10 -> 73.21 | Helped |
| Set re-voxelization scale factor alpha to 0.15 rather than 0.20 (ablation) | CAGroup3D class-aware grouping · ScanNet V2 val | mAP@0.5: 58.77 -> 60.31 | Helped |
| Lower semantic threshold from 0.20 to 0.06 (ablation) | CAGroup3D class-aware grouping · ScanNet V2 val | mAP@0.5: 59.50 -> 60.27 | Helped |
| Increase RoI abstraction depth from two blocks to three (ablation) | CAGroup3D RoI module · ScanNet V2 val | mAP@0.25: 74.50 -> 73.91 | Hurt |
| Increase first-layer proposal sampling resolution to 11 × 11 × 11 (ablation) | CAGroup3D RoI module · ScanNet V2 val | mAP@0.25: 74.50 -> 74.44 | No clear effect |
Closest papers by meaning
- SPGroup3D: Superpoint Grouping Network for Indoor 3D Object Detection (AAAI 2024 · similarity 0.89)
- LION: Linear Group RNN for 3D Object Detection in Point Clouds (NeurIPS 2024 · similarity 0.88)
- Voxel Mamba: Group-Free State Space Models for Point Cloud based 3D Object Detection (NeurIPS 2024 · similarity 0.88)
- ImOV3D: Learning Open Vocabulary Point Clouds 3D Object Detection from Only 2D Images (NeurIPS 2024 · similarity 0.87)
- Towards Learning Group-Equivariant Features for Domain Adaptive 3D Detection (NeurIPS 2024 · similarity 0.87)
- SP3D: Boosting Sparsely-Supervised 3D Object Detection via Accurate Cross-Modal Semantic Prompts (CVPR 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.