Research map / Point clouds / Segmentation and detection
Deep Hough Voting for 3D Object Detection in Point Clouds
Charles R. Qi, Or Litany, Kaiming He, Leonidas J. Guibas
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 voting (vote then cluster) instead of direct box proposal from seeds (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 53.0 -> 57.7 | Improved |
| add voting instead of direct box proposal from seeds (ablation) | VoteNet · ScanNetV2 val | mAP@0.25: 45.4 -> 58.6 | Improved |
| use learned PointNet vote aggregation with max pooling instead of manual pooling (max, avg, RBF) (ablation) | VoteNet · SUN RGB-D val | mAP@0.25 | Improved |
| increase aggregation radius beyond about 0.2 (ablation) | VoteNet · SUN RGB-D val | mAP@0.25 | Hurt |
| vote from deeper seed features (SA2 to FP2) (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 51.2 -> 57.7 | Improved |
| extend seed context to FP3 beyond FP2 (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 57.7 -> 57.1 | No clear effect |
| generate 3 votes per seed instead of 1 (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 57.7 -> 55.8 | Hurt |
| add random number to break symmetry with 3 votes per seed (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 55.8 -> 55.8 | No clear effect |
| change proposal sampling from seed FPS to random or vote FPS (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 57.7 -> 57.5 | No clear effect |
| add the height feature (ablation) | VoteNet · SUN RGB-D val | mAP@0.25: 57.0 -> 57.7 | Improved |
| add the height feature (ablation) | VoteNet · ScanNetV2 val | mAP@0.25: 58.1 -> 58.6 | Improved |
| use point-based voting with geometry only instead of 3D CNN 3D-SIS (geo only) | 3D-SIS · ScanNetV2 val | mAP@0.25: 25.4 -> 58.6 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- GSPN — GSPN: Generative Shape Proposal Network for 3D Instance Segmentation in Point Cloud (CVPR 2019) · better on 2 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
- Back-Tracing Representative Points for Voting-Based 3D Object Detection in Point Clouds (CVPR 2021 · similarity 0.92)
- ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020 · similarity 0.92)
- SPOT: Selective Point Cloud Voting for Better Proposal in Point Cloud Object Detection (ECCV 2020 · similarity 0.91)
- Prototypical VoteNet for Few-Shot 3D Point Cloud Object Detection (NeurIPS 2022 · similarity 0.90)
- VENet: Voting Enhancement Network for 3D Object Detection (ICCV 2021 · similarity 0.90)
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection (CVPR 2018 · similarity 0.90)
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.
