Research map / Point clouds / Segmentation and detection
SPOT: Selective Point Cloud Voting for Better Proposal in Point Cloud Object Detection
Hongyuan Du, Linjun Li, Bo Liu, Nuno Vasconcelos
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 |
|---|---|---|---|
| replace VoteNet regression with SPOT selective voting | VoteNet · ScanNet V2 val | mAP@0.5: 33.5 -> 40.4 | Improved |
| replace VoteNet regression with SPOT selective voting (ablation) | VoteNet · ScanNet V2 val | AR@0.5: 53.3 -> 57.8 | Improved |
| keep best 1 vote per interest point instead of single regression (ablation) | VoteNet · ScanNet V2 val | mAP@0.5: 33.5 -> 38.7 | Improved |
| add dynamic suspicious-coincidence selection instead of best 1 per interest point (ablation) | VoteNet+SPOT · ScanNet V2 val | mAP@0.5: 38.7 -> 40.4 | Improved |
| keep best 2 votes per interest point instead of best 1 (ablation) | VoteNet · ScanNet V2 val | mAP@0.5: 38.7 -> 38.4 | Hurt |
| keep best 3 votes per interest point instead of best 1 (ablation) | VoteNet · ScanNet V2 val | mAP@0.5: 38.7 -> 37.1 | Hurt |
| increase number of votes up to 1,024 (ablation) | SPOT · ScanNet V2 val | mAP@0.5 | Mixed |
| use 24 sub-regions instead of too few or too many (ablation) | SPOT · ScanNet V2 val | mAP@0.5 | Mixed |
| add SPOT to PointRCNN | PointRCNN · ScanNet V2 val | mAP@0.5: 25.4 -> 27.8 | Improved |
| add SPOT to PointRCNN-SC | PointRCNN-SC · SUN RGB-D val | mAP@0.5: 24.5 -> 27.7 | Improved |
| add SPOT to VoteNet | VoteNet · SUN RGB-D val | mAP@0.5: 32.3 -> 36.3 | Improved |
| add SPOT to PointRCNN-SC | PointRCNN-SC · ScanNet V2 val | mAP@0.25: 57.0 -> 57.4 | Improved |
Closest papers by meaning
- Deep Hough Voting for 3D Object Detection in Point Clouds (ICCV 2019 · similarity 0.91)
- Back-Tracing Representative Points for Voting-Based 3D Object Detection in Point Clouds (CVPR 2021 · similarity 0.89)
- ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020 · similarity 0.88)
- Not All Points Are Equal: Learning Highly Efficient Point-Based Detectors for 3D LiDAR Point Clouds (CVPR 2022 · similarity 0.88)
- Prototypical VoteNet for Few-Shot 3D Point Cloud Object Detection (NeurIPS 2022 · similarity 0.87)
- AS-Det: Active Sampling for Adaptive 3D Object Detection in Point Clouds (AAAI 2025 · similarity 0.86)
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.
