Research map / 3D vision / Depth and stereo
Accurate 3-DoF Camera Geo-Localization via Ground-to-Satellite Image Matching
Yujiao Shi, Xin Yu, Liu Liu, Dylan Campbell, Piotr Koniusz, Hongdong Li
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 the projective transform branch to the polar-only pipeline (ablation) | coarse localization · CVUSA FoV=180 orientation unknown | r@1: 48.53 -> 54.27 | Improved |
| add the projective transform branch (ablation) | coarse localization · CVACT_val FoV=180 orientation unknown | r@1: 49.12 -> 52.98 | Improved |
| add the projective transform branch (ablation) | coarse localization · CVUSA FoV=90 orientation unknown | r@1: 16.19 -> 11.30 | Hurt |
| add the projective transform branch (ablation) | coarse localization · CVACT_val FoV=70 orientation unknown | r@1: 8.29 -> 4.60 | Hurt |
| add the polar transform branch (ablation) | coarse localization · CVUSA FoV=360 orientation unknown | r@1: 65.33 -> 78.94 | Improved |
| add the polar transform branch (ablation) | coarse localization · CVACT_val FoV=180 orientation unknown | r@1: 27.95 -> 52.98 | Improved |
| use both polar and projective transforms for orientation estimation instead of projective only (ablation) | orientation estimation · CVUSA FoV=180 | r@1: 52.00 -> 54.27 | Improved |
| use both transforms for orientation estimation instead of polar only (ablation) | orientation estimation · CVUSA FoV=180 | r@1: 53.88 -> 54.27 | Improved |
| align orientation only to the matching satellite image instead of the whole database (ablation) | DSM orientation alignment · CVUSA FoV=180 | r@1: 54.27 -> 79.44 | Improved |
| remove orientation augmentation knowledge: unknown orientation versus aligned | Ours · CVUSA FoV=180 | r@1: 75.65 -> 54.27 | Hurt |
| replace SAFA spatial-aware modules with new projective geometry (this work vs SAFA) | Ours · CVACT_test | top-1 recall: 55.50% -> 60.46% | Improved |
| random orientation augmentation in training | CVFT and SAFA · CVUSA FoV=360 | r@1: 61.43 -> 23.13 (CVFT); 89.84 -> 53.33 (SAFA) | Hurt |
Closest papers by meaning
- Beyond Cross-View Image Retrieval: Highly Accurate Vehicle Localization Using Satellite Image (CVPR 2022 · similarity 0.91)
- Boosting 3-DoF Ground-to-Satellite Camera Localization Accuracy via Geometry-Guided Cross-View Transformer (ICCV 2023 · similarity 0.90)
- Image-Based Geo-Localization Using Satellite Imagery (IJCV 2019 · similarity 0.90)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.87)
- Scaling Image Geo-Localization to Continent Level (NeurIPS 2025 · similarity 0.87)
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.87)
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.
