Research map / Label-efficient and robust learning / Recognition and long-tail
Spatial Retrieval Augmented Autonomous Driving
Xiaosong Jia, Chenhe Zhang, Yule Jiang, Songbur Wong, Zhiyuan Zhang, Chen Chen, Shaofeng Zhang, Xuanhe Zhou, Xue Yang, Junchi Yan, Yu-Gang Jiang
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 |
|---|---|---|---|
| add geographic image retrieval to online mapping | MapTR (110 epochs) · nuScenes-Geography | mAP: 59.3 -> 72.7 | Improved |
| add geographic image retrieval to online mapping | MapTRv2 (110 epochs) · nuScenes-Geography | mAP: 68.7 -> 78.2 | Improved |
| add geographic images to occupancy prediction | FBOcc · Occ3D-nuScenes | overall mIoU: 39.11 -> 39.74 | Improved |
| add geographic images to occupancy prediction | FlashOCC:M1 · Occ3D-nuScenes | overall mIoU: 31.92 -> 32.35 | Improved |
| add geographic images to 3D object detection | BEVDet · nuScenes detection | mAP: 30.85 -> 30.69 | No clear effect |
| add geographic images to 3D object detection | BEVFormer · nuScenes detection | NDS: 51.70 -> 51.80 | No clear effect |
| add geographic images to end-to-end planning | VAD · nuScenes full val | avg L2 (m): 0.64 -> 0.64 | No clear effect |
| add geographic images to end-to-end planning | VAD · nuScenes night subset | avg collision (%): 0.55 -> 0.48 | Improved |
| add geographic images to generative world model | UniMLVG · nuScenes | FVD: 36.10 -> 29.97 | Improved |
| add geographic images to generative world model | MagicDriveDiT · nuScenes | FVD: 84.43 -> 81.52 | Improved |
| add geographic images (vs none) (ablation) | FlashOcc / UniMLVG · Ablation Table 7a | static mIoU / FVD: 46.66 -> 47.86 mIoU; 35.42 -> 29.97 FVD | Improved |
| add 3D positional encoding for geographic patches (ablation) | FlashOcc / UniMLVG · Ablation Table 7b | static mIoU / FVD: 46.22 -> 47.86 mIoU; 32.82 -> 29.97 FVD | Improved |
Closest papers by meaning
- PriorEye: Geospatial Visual Priors for End-to-End Autonomous Driving (ECCV 2026 · similarity 0.88)
- The Road Ahead in Autonomous Driving: The KITScenes Multimodal Dataset (NeurIPS 2026 · similarity 0.87)
- Online Map Vectorization for Autonomous Driving: A Rasterization Perspective (NeurIPS 2023 · similarity 0.87)
- Pix2map: Cross-Modal Retrieval for Inferring Street Maps From Images (CVPR 2023 · similarity 0.87)
- Ithaca365: Dataset and Driving Perception Under Repeated and Challenging Weather Conditions (CVPR 2022 · similarity 0.87)
- SpaceDrive: Infusing Spatial Awareness into VLM-based Autonomous Driving (CVPR 2026 · similarity 0.86)
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.
