Research map / Point clouds / Segmentation and detection
PointRCNN: 3D Object Proposal Generation and Detection From Point Cloud
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 PointRCNN instead of the strongest listed baseline | KITTI car test detection · KITTI test, Car Moderate, IoU=0.7 | AP: 73.66 -> 75.76 | Helped |
| use PointRCNN instead of the strongest listed baseline | KITTI cyclist test detection · KITTI test, Cyclist Moderate, IoU=0.5 | AP: 56.77 -> 59.60 | Helped |
| use PointRCNN instead of the strongest listed baseline | KITTI pedestrian test detection · KITTI test, Pedestrian Moderate, IoU=0.5 | AP: 44.89 -> 41.78 | Hurt |
| add GT-AUG (ablation) | PointRCNN · KITTI val, Car Moderate | AP: 77.67 -> 78.63 | Helped |
| add canonical transformation (ablation) | PointRCNN refinement network · KITTI val, Car Moderate | AP: 13.68 -> 77.67 | Helped |
| add stage-1 point features (ablation) | PointRCNN refinement network · KITTI val, Car Moderate | AP: 74.96 -> 77.67 | Helped |
| add camera-depth feature (ablation) | PointRCNN refinement network · KITTI val, Car Moderate | AP: 76.79 -> 77.67 | Helped |
| add segmentation-mask feature (ablation) | PointRCNN refinement network · KITTI val, Car Moderate | AP: 76.64 -> 77.67 | Helped |
| use 1.0m context width instead of no context (ablation) | PointRCNN point-cloud pooling · KITTI val, Car Hard | AP: 68.92 -> 76.30 | Helped |
| set context width to 2.0m rather than 1.0m (ablation) | PointRCNN point-cloud pooling · KITTI val, Car Moderate | AP: 77.67 -> 76.61 | Hurt |
| use full bin-based regression loss instead of alternative regression losses (ablation) | PointRCNN stage-1 proposal generation · KITTI val, Car proposal generation | recall | Helped |
| use residual-cos-based loss instead of residual-based loss (ablation) | PointRCNN stage-1 proposal generation · KITTI val, Car proposal generation | recall | Helped |
Closest papers by meaning
- PTNET: A PROPOSAL-CENTRIC TRANSFORMER NET- WORK FOR 3D OBJECT DETECTION (ICLR 2026 · similarity 0.91)
- VoxelNet: End-to-End Learning for Point Cloud Based 3D Object Detection (CVPR 2018 · similarity 0.89)
- Point-GNN: Graph Neural Network for 3D Object Detection in a Point Cloud (CVPR 2020 · similarity 0.89)
- Frustum PointNets for 3D Object Detection From RGB-D Data (CVPR 2018 · similarity 0.89)
- PointCNN++: Performant Convolution on Native Points (CVPR 2026 · similarity 0.88)
- LinNet: Linear Network for Efficient Point Cloud Representation Learning (NeurIPS 2024 · 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.