Research map / 3D vision / Depth and stereo
OpenCVL: An Open, Diverse, and Large-Scale Dataset for Fine-Grained Cross-View Localization
Zimin Xia, Mubariz Zaffar, Junsheng Fu, Alexandre ALahi, Julian Kooij
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 noisy Mapillary data to ZOD training (ablation) | Loc2 · OpenCVL in-the-wild test | mean localization error (m): 8.61 -> 7.90 | Improved |
| downweight Mapillary loss by 0.3 (ablation) | Loc2 · OpenCVL cross-area test | mean localization error (m): 6.87 -> 6.72 | Improved |
| downweight Mapillary loss by 0.3 (ablation) | Loc2 · OpenCVL in-the-wild test | mean orientation error (deg): 25.87 -> 26.07 | No clear effect |
| add Mapillary data (weighted) to ZOD training (ablation) | Loc2 · OpenCVL snowy test | mean orientation error (deg): 8.87 -> 5.09 (median 3.26 -> 3.52) | Mixed |
| train on OpenCVL instead of KITTI | Loc2 · OpenCVL cross-area test | mean localization error (m): 15.57 -> 6.72 | Improved |
| use orientation prior during KITTI training (ablation) | Loc2 · OpenCVL cross-area test | mean localization error (m): 15.57 -> 20.07 | Hurt |
| train on OpenCVL instead of ZOD only (ablation) | Loc2 · KITTI cross-area (no prior) | mean localization error (m): 12.04 -> 11.25 | Improved |
| train on OpenCVL instead of KITTI | Loc2 · KITTI cross-area (no prior) | mean orientation error (deg): 55.18 -> 71.17 | Hurt |
| use corrected Mapillary poses instead of raw GNSS (ablation) | Loc2 finetuning · OpenCVL in-the-wild | mean localization error (m): 7.41 -> 6.92 | Improved |
| use OpenSfM poses instead of raw GNSS (ablation) | Loc2 finetuning · OpenCVL in-the-wild | mean orientation error (deg): 22.09 -> 23.44 | Mixed |
| joint training OpenCVL + KITTI | Loc2 · KITTI cross-area (with prior) | mean orientation error (deg): 3.32 -> 4.06 | Hurt |
| pretrain on OpenCVL then finetune on KITTI | Loc2 · KITTI same-area (with prior) | mean localization error (m): 1.13 -> 0.93 | Improved |
Closest papers by meaning
- Sector-Level Cross-View Geo-Localization with Implicit Orientation via Azimuthal Scanning (ECCV 2026 · similarity 0.91)
- Unleashing Unlabeled Data: A Paradigm for Cross-View Geo-Localization (CVPR 2024 · similarity 0.89)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.88)
- Weather-Robust Cross-View Geo-Localization via Prototype-Based Semantic Part Discovery (NeurIPS 2026 · similarity 0.88)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.88)
- 360Loc: A Dataset and Benchmark for Omnidirectional Visual Localization with Cross-device Queries (CVPR 2024 · similarity 0.88)
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.
