Research map / Detection and segmentation / 3D detection and driving
Learning to Evaluate Perception Models Using Planner-Centric Metrics
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 PKL rather than NDS to rank which detection errors are more dangerous | 3D object detection on nuScenes · Human evaluation | share of responses preferring the danger ranking: 21% -> 79% | Helped |
| Add training on trajectories of all annotated vehicles instead of ego-only trajectories (ablation) | Planner · nuScenes val | Top 5 accuracy: 35.28 -> 37.41% | Helped |
| Add training on trajectories of all annotated vehicles instead of ego-only trajectories (ablation) | Planner · nuScenes val | Top 1 accuracy: 17.18 -> 19.39% | Helped |
| Add training on trajectories of all annotated vehicles instead of ego-only trajectories (ablation) | Planner · nuScenes val | L2 distance to ground-truth future position: 1.47 -> 1.27 m | Helped |
| Add loss clipping (ablation) | Planner · nuScenes val | Top 5 accuracy: 35.72 -> 37.41% | Helped |
| Add loss clipping (ablation) | Planner · nuScenes val | Top 1 accuracy: 18.33 -> 19.39% | Helped |
| Add loss clipping (ablation) | Planner · nuScenes val | L2 distance to ground-truth future position: 1.45 -> 1.27 m | Helped |
| Add positive-example weighting (ablation) | Planner · nuScenes val | Top 5 accuracy: 35.47 -> 37.41% | Helped |
| Add positive-example weighting (ablation) | Planner · nuScenes val | Top 1 accuracy: 18.74 -> 19.39% | Helped |
| Add positive-example weighting (ablation) | Planner · nuScenes val | L2 distance to ground-truth future position: 1.42 -> 1.27 m | Helped |
| Add dropout (ablation) | Planner · nuScenes val | Top 5 accuracy: 34.25 -> 37.41% | Helped |
| Add dropout (ablation) | Planner · nuScenes val | Top 1 accuracy: 18.59 -> 19.39% | Helped |
Closest papers by meaning
- Size-invariance Matters: Rethinking Metrics and Losses for Imbalanced Multi-object Salient Object Detection (ICML 2024 · similarity 0.86)
- OpenAD: Open-World Autonomous Driving Benchmark for 3D Object Detection (NeurIPS 2025 · similarity 0.86)
- RCP-Bench: Benchmarking Robustness for Collaborative Perception Under Diverse Corruptions (CVPR 2025 · similarity 0.85)
- DriveGEN: Generalized and Robust 3D Detection in Driving via Controllable Text-to-Image Diffusion Generation (CVPR 2025 · similarity 0.85)
- AIDE: An Automatic Data Engine for Object Detection in Autonomous Driving (CVPR 2024 · similarity 0.85)
- CorrBEV: Multi-View 3D Object Detection by Correlation Learning with Multi-modal Prototypes (CVPR 2025 · similarity 0.84)
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.