Research map / Detection and segmentation / 3D detection and driving
Learnability-Driven Submodular Optimization for Active Roadside 3D Detection
Ruiyu Mao, Baoming Zhang, Nicholas Ruozzi, Yunhui Guo
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 |
|---|---|---|---|
| select by learnability (LH3D) instead of PPAL | LH3D · DAIR-V2X-I (BEVHeight) | average AP3D hard: 36.41 -> 38.50 | Improved |
| LH3D instead of HUA | LH3D · DAIR-V2X-I (BEVHeight) | average AP3D moderate: 32.81 -> 38.33 | Improved |
| LH3D instead of RANDOM | LH3D · DAIR-V2X-I (BEVHeight) | average AP3D moderate: 34.43 -> 38.33 | Improved |
| LH3D instead of PPAL on BEVSpread backbone | LH3D · DAIR-V2X-I (BEVSpread) | average AP3D hard: 33.59 -> 36.07 | Improved |
| LH3D instead of PPAL on BEVDet backbone | LH3D · DAIR-V2X-I (BEVDet) | average AP3D easy: 31.85 -> 32.63 | Improved |
| LH3D instead of PPAL on Rope3D | LH3D · Rope3D (BEVHeight) | average AP3D easy: 14.73 -> 14.97 | No clear effect |
| LH3D instead of PPAL on Rope3D | LH3D · Rope3D (BEVHeight) | average AP3D moderate: 14.05 -> 15.77 | Improved |
| LH3D vs PPAL on pedestrians | LH3D · DAIR-V2X-I (BEVHeight) | Pedestrian AP3D moderate: 18.47 -> 17.50 | Hurt |
| train on ambiguous split instead of learnable split (ablation) | detector · DAIR-V2X-I human study | AP cars and pedestrians | Hurt |
| LH3D selection vs baselines on class balance | LH3D · DAIR-V2X-I | class diversity entropy | Improved |
| increase initial labeled set from 100 to 500 images (ablation) | LH3D · DAIR-V2X-I (hard) | AP3D: gap narrows to about 2.5% | Mixed |
Closest papers by meaning
- Rope3D: The Roadside Perception Dataset for Autonomous Driving and Monocular 3D Object Detection Task (CVPR 2022 · similarity 0.89)
- RECO: Region-Aware Compensation for Extrinsic Perturbations in Roadside 3D Detection (ECCV 2026 · similarity 0.88)
- BEVSpread: Spread Voxel Pooling for Bird's-Eye-View Representation in Vision-based Roadside 3D Object Detection (CVPR 2024 · similarity 0.87)
- RoadSceneBench: A Lightweight Benchmark for Mid-Level Road Scene Understanding (CVPR 2026 · similarity 0.86)
- Uncertainty Meets Diversity: A Comprehensive Active Learning Framework for Indoor 3D Object Detection (CVPR 2025 · similarity 0.86)
- 3D StreetUnveiler with Semantic-aware 2DGS - a simple baseline (ICLR 2025 · 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.
