Research map / Detection and segmentation / 3D detection and driving
RangeIoUDet: Range Image Based Real-Time 3D Object Detector Optimized by Intersection Over Union
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 RangeIoUDet instead of the strongest listed single-stage baseline | KITTI 3D detection · KITTI test, Cyclist moderate | AP R40: 65.07 -> 67.77 | Helped |
| Use RangeIoUDet instead of PV-RCNN | KITTI 3D detection · KITTI test, Car moderate | AP R40: 81.43 -> 79.80 | Hurt |
| Use RangeIoUDet instead of PV-RCNN | Self-built 3D detection · Car Height>40 | AP R40: 98.81 -> 99.09 | Helped |
| Use RangeIoUDet instead of PV-RCNN | Self-built 3D detection · Pedestrian Height>40 | AP R40: 56.21 -> 53.38 | Hurt |
| Add 3D IoU box regression instead of Smooth L1 alone (ablation) | RangeIoUDet · KITTI validation, Car easy | AP R40: 87.79 -> 88.49 | Helped |
| Use 3D GIoU loss rather than 3D IoU loss (ablation) | RangeIoUDet · KITTI validation, Car easy | AP R40: 88.49 -> 89.61 | Helped |
| Combine Smooth L1 center supervision with 3D GIoU box regression (ablation) | RangeIoUDet · KITTI validation, Car easy | AP R40: 89.61 -> 90.58 | Helped |
| Add 3D IoU quality-score supervision (ablation) | RangeIoUDet with HyGIoUreg · KITTI validation, Car easy | AP R40: 90.58 -> 90.59 | Helped |
| Multiply classification confidence by predicted quality score (ablation) | RangeIoUDet with box quality supervision · KITTI validation, Car easy | AP R40: 90.59 -> 91.10 | Helped |
| Add point-based IoU supervision using a fully connected layer (ablation) | RangeIoUDet · KITTI validation, Car moderate | AP R40: 82.42 -> 82.34 | No clear effect |
| Add local PointNet to use the 3D receptive field for point-based IoU supervision (ablation) | RangeIoUDet · KITTI validation, Car moderate | AP R40: 82.42 -> 82.99 | Helped |
| Use parallel rather than cascade multi-scale local PointNet layers (ablation) | RangeIoUDet · KITTI validation, Car moderate | AP R40: 82.77 -> 83.19 | Helped |
Closest papers by meaning
- Fully Convolutional One-Stage 3D Object Detection on LiDAR Range Images (NeurIPS 2022 · similarity 0.92)
- LaserNet: An Efficient Probabilistic 3D Object Detector for Autonomous Driving (CVPR 2019 · similarity 0.91)
- RangePerception: Taming LiDAR Range View for Efficient and Accurate 3D Object Detection (NeurIPS 2023 · similarity 0.90)
- Pseudo-LiDAR From Visual Depth Estimation: Bridging the Gap in 3D Object Detection for Autonomous Driving (CVPR 2019 · similarity 0.89)
- BEV-MAE: Bird’s Eye View Masked Autoencoders for Point Cloud Pre-training in Autonomous Driving Scenarios (AAAI 2024 · similarity 0.87)
- STONE: A Submodular Optimization Framework for Active 3D Object Detection (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.