Research map / Point clouds / Segmentation and detection
LMD: Light-Weight Prediction Quality Estimation for Object Detection in Lidar Point Clouds
Tobias Riedlinger, Marius Schubert, Sarina Penquitt, Jan-Marcel Kezmann, Pascal Colling, Karsten Kahl, Lutz Roese-Koerner, Michael Arnold, Urs Zimmermann, Matthias Rottmann
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 |
|---|---|---|---|
| use all 90 LMD features instead of score only | gradient boosting meta classifier · nuScenes CenterPoint | AUROC: 0.8623 -> 0.9628 | Improved |
| add Lidar and proposal statistics beyond box features | gradient boosting meta classifier · nuScenes CenterPoint | AUROC: 0.9529 -> 0.9628 | Improved |
| add box features to score | gradient boosting meta regressor · nuScenes CenterPoint | R^2: 0.4732 -> 0.6792 | Improved |
| add all LMD features to box features | gradient boosting meta regressor · nuScenes PointPillars | R^2: 0.5593 -> 0.6413 | Improved |
| add all LMD features to box features | gradient boosting meta regressor · KITTI PointPillars | R^2: 0.7131 -> 0.7287 | Improved |
| use non-linear meta classifier (random forest) instead of logistic regression (ablation) | LMD features · nuScenes CenterPoint | AUROC: 0.9450 -> 0.9581 | Improved |
| use gradient boosting instead of logistic regression (ablation) | LMD features · nuScenes CenterPoint | AUROC: 0.9450 -> 0.9628 | Improved |
| use meta model choice on score only (ablation) | score baseline · nuScenes CenterPoint | AUROC: 0.8644 -> 0.8617 | No clear effect |
| use gradient boosting instead of ridge regression (ablation) | LMD features · nuScenes CenterPoint | R^2: 0.6451 -> 0.7296 | Improved |
| select only five greedy features instead of all 90 (ablation) | LMD · nuScenes CenterPoint | AUROC: 0.19 pp worse | No clear effect |
| meta classifier confidence instead of objectness score | detector confidence · nuScenes CenterPoint | ECE: 8.07 pp | Improved |
| rank false positives by LMD predicted IoU instead of score | annotation error proposals · nuScenes | annotation errors in top 100: 6 -> 43 | Improved |
Closest papers by meaning
- LiDAR Distillation: Bridging the Beam-Induced Domain Gap for 3D Object Detection (ECCV 2022 · similarity 0.87)
- LiDAR-PTQ: Post-Training Quantization for Point Cloud 3D Object Detection (ICLR 2024 · similarity 0.87)
- LiON: Learning Point-Wise Abstaining Penalty for LiDAR Outlier DetectioN Using Diverse Synthetic Data (AAAI 2025 · similarity 0.87)
- An Empirical Study of the Generalization Ability of Lidar 3D Object Detectors to Unseen Domains (CVPR 2024 · similarity 0.86)
- GBlobs: Explicit Local Structure via Gaussian Blobs for Improved Cross-Domain LiDAR-based 3D Object Detection (CVPR 2025 · similarity 0.86)
- MOS: Model Synergy for Test-Time Adaptation on LiDAR-Based 3D Object Detection (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.
