Research map / Point clouds / Segmentation and detection
Back-Tracing Representative Points for Voting-Based 3D Object Detection in Point Clouds
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 BRNet’s generative representative-point back-tracing, seed revisiting, and refinement | VoteNet-style 3D detector · ScanNet V2 val | mAP@0.50: 48.1 -> 50.9 | Helped |
| Use BRNet’s generative representative-point back-tracing, seed revisiting, and refinement | VoteNet-style 3D detector · SUN RGB-D V1 val | mAP@0.50: 39.0 -> 43.7 | Helped |
| Use class-agnostic bounding-box regression (ablation) | VoteNet · ScanNet V2 | mAP@0.50: 33.5 -> 40.8 | Helped |
| Use class-agnostic bounding-box regression (ablation) | VoteNet · SUN RGB-D | mAP@0.50: 32.9 -> 37.6 | Helped |
| Fuse vote features with corresponding seed-point features (ablation) | VoteNet · ScanNet V2 | mAP@0.50: 33.5 -> 37.6 | Helped |
| Fuse vote features with corresponding seed-point features (ablation) | VoteNet · SUN RGB-D | mAP@0.50: 32.9 -> 33.6 | Helped |
| Use 6 rather than the default 12 Ray-sampled representative points (ablation) | BRNet · ScanNet V2 | mAP@0.50: 50.9 -> 48.3 | Hurt |
| Use 18 rather than the default 12 Ray-sampled representative points (ablation) | BRNet · ScanNet V2 | mAP@0.50: 50.9 -> 48.4 | Hurt |
| Use Grid sampling with 8 points instead of Ray sampling with 12 points (ablation) | BRNet · ScanNet V2 | mAP@0.50: 50.9 -> 49.1 | Hurt |
| Use Grid sampling with 27 points instead of Ray sampling with 12 points (ablation) | BRNet · ScanNet V2 | mAP@0.50: 50.9 -> 49.2 | Hurt |
Closest papers by meaning
- Prototypical VoteNet for Few-Shot 3D Point Cloud Object Detection (NeurIPS 2022 · similarity 0.89)
- ImVoteNet: Boosting 3D Object Detection in Point Clouds With Image Votes (CVPR 2020 · similarity 0.88)
- PointRCNN: 3D Object Proposal Generation and Detection From Point Cloud (CVPR 2019 · similarity 0.87)
- Canonical Voting: Towards Robust Oriented Bounding Box Detection in 3D Scenes (CVPR 2022 · similarity 0.86)
- PointCFormer: A Relation-Based Progressive Feature Extraction Network for Point Cloud Completion (AAAI 2025 · similarity 0.85)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · similarity 0.85)
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.