Research map / Point clouds / Segmentation and detection
PointPainting: Sequential Fusion for 3D Object Detection
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 |
|---|---|---|---|
| Add projected image segmentation scores to PointPillars lidar points | PointPillars · KITTI val, BEV, moderate | mAP: 73.78 -> 76.27 | Helped |
| Add projected image segmentation scores to VoxelNet lidar points | VoxelNet · KITTI val, BEV, moderate | mAP: 71.83 -> 73.55 | Helped |
| Add projected image segmentation scores to PointRCNN lidar points | PointRCNN · KITTI val, BEV, moderate | mAP: 72.42 -> 75.80 | Helped |
| Add projected image segmentation scores to PointRCNN lidar points | PointRCNN · KITTI test, BEV | mAP: 66.92 -> 69.86 | Helped |
| Add projected image segmentation scores to PointPillars+ lidar points | PointPillars+ · nuScenes test | mAP: 40.1 -> 46.4 | Helped |
| Add projected image segmentation scores to PointPillars+ lidar points | PointPillars+ · nuScenes test | NDS: 55.0 -> 58.1 | Helped |
| Use ground-truth 3D boxes rather than predicted segmentation to paint the points (ablation) | PointPillars · nuScenes val, quarter training | mAP: +27 mAP | Helped |
| Replace segmentation scores with one-hot class labels (ablation) | Painted PointPillars · nuScenes val, quarter training | NDS: unchanged | No clear effect |
| Replace segmentation scores with one-hot class labels (ablation) | Painted PointPillars · nuScenes val, quarter training | mAP: +0.4 | No clear effect |
| Use segmentation from the previous image for pipelined consecutive matching (ablation) | Painted PointPillars · nuScenes val, quarter training | mAP: 33.9 -> 33.9 | No clear effect |
| Use segmentation from the previous image for pipelined consecutive matching (ablation) | Painted PointPillars · nuScenes val, quarter training | NDS: 46.3 -> 46.4 | No clear effect |
Closest papers by meaning
- PointFusion: Deep Sensor Fusion for 3D Bounding Box Estimation (CVPR 2018 · similarity 0.89)
- Adaptive-Smooth LiDAR-Camera Knowledge Distillation with Heterogeneous Fusion for Multi-View 3D Object Detection (AAAI 2026 · similarity 0.88)
- PointAugmenting: Cross-Modal Augmentation for 3D Object Detection (CVPR 2021 · similarity 0.88)
- Pre-training LiDAR-based 3D Object Detectors through Colorization (ICLR 2024 · similarity 0.88)
- Towards Unsupervised Object Detection From LiDAR Point Clouds (CVPR 2023 · similarity 0.88)
- ALSO: Automotive Lidar Self-Supervision by Occupancy Estimation (CVPR 2023 · similarity 0.88)
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.