Research map / Detection and segmentation / 3D detection and driving
Correlation Field for Boosting 3D Object Detection in Structured Scenes
Jianhua Sun, Hao-Shu Fang, Xianghui Zhu, Jiefeng Li, Cewu Lu
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 |
|---|---|---|---|
| Apply both CorrelaBoost strategies instead of vanilla augmentation (ablation) | VoteNet · ScanNetV2 | mAP@0.5: 33.5 -> 41.2 | Improved |
| Apply both CorrelaBoost strategies instead of vanilla augmentation (ablation) | VoteNet · SUN RGB-D | mAP@0.5: 32.7 -> 37.7 | Improved |
| Add Category Consistent Exchanging (ablation) | VoteNet · ScanNetV2 | mAP@0.5: 33.5 -> 36.9 | Improved |
| Add Energy Optimized Transformation (ablation) | VoteNet · ScanNetV2 | mAP@0.5: 33.5 -> 38.8 | Improved |
| Add Category Consistent Exchanging (ablation) | VoteNet · SUN RGB-D | mAP@0.5: 32.7 -> 35.1 | Improved |
| Add Energy Optimized Transformation (ablation) | VoteNet · SUN RGB-D | mAP@0.5: 32.7 -> 36.3 | Improved |
| Use correlation-field-guided pasting rather than random paste | VoteNet · ScanNetV2 | mAP@0.5: 34.2 -> 41.2 | Improved |
| Use correlation-field-guided pasting rather than random paste | VoteNet · SUN RGB-D | mAP@0.5: 33.6 -> 37.7 | Improved |
| Guide same-category exchanges with shape and orientation similarity rather than random pairing (ablation) | VoteNet · ScanNetV2 | AP50: 35.5 -> 36.9 | Improved |
| Vary the shape and orientation similarity weights across the reported settings (ablation) | VoteNet with exchanging only · ScanNetV2 | AP50: 35.5; 36.3; 36.7; 36.9; 36.6; 36.1 | Mixed |
| Vary correlation-field attenuation index gamma across the reported settings (ablation) | VoteNet with transformation only · ScanNetV2 | AP50: 38.8; 38.6; 38.7 | Mixed |
| Vary correlation-field constant k across the reported settings (ablation) | VoteNet with transformation only · ScanNetV2 | AP50: 38.4; 38.6; 38.3; 38.8 | Mixed |
Closest papers by meaning
- LiDAR-Aug: A General Rendering-Based Augmentation Framework for 3D Object Detection (CVPR 2021 · similarity 0.86)
- Improving 3D Object Detection through Progressive Population Based Augmentation (ECCV 2020 · similarity 0.86)
- AGO-Net: Association-Guided 3D Point Cloud Object Detection Network (TPAMI 2021 · similarity 0.86)
- Rethinking Few-shot 3D Point Cloud Semantic Segmentation (CVPR 2024 · similarity 0.85)
- Cubify Anything: Scaling Indoor 3D Object Detection (CVPR 2025 · similarity 0.85)
- Leveraging Imagery Data with Spatial Point Prior for Weakly Semi-supervised 3D Object Detection (AAAI 2024 · similarity 0.84)
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.
