Research map / 3D vision / Depth and stereo
VGA: Empowering Aerial-Ground Localization by Visual Geometry Alignment
Tao Jun Lin, Yujiao Shi, 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 gravity prior in inference-time optimization (ablation) | MASt3R (AerialMegaDepth+MatrixCity) · MatrixCity (BigCity) | AUC@30: 29.12 -> 33.38 | Improved |
| add planar (BEV) alignment prior alone (ablation) | MASt3R (AerialMegaDepth+MatrixCity) · MatrixCity (BigCity) | AUC@30: 29.12 -> 31.36 | Improved |
| add planar alignment on top of gravity prior (ablation) | VGA · MatrixCity (BigCity) | AUC@30: 33.38 -> 34.97 | Improved |
| add joint gravity and planar optimization (ablation) | MASt3R (AerialMegaDepth+MatrixCity) · MatrixCity (BigCity) | RTA@25: 37.10 -> 45.64 | Improved |
| add joint gravity and planar optimization (ablation) | MASt3R (AerialMegaDepth+MatrixCity) · MatrixCity (BigCity) | RRA@25: 33.54 -> 46.70 | Improved |
| add joint optimization (ablation) | MASt3R · MatrixCity (BigCity) | RTA@2: 15.30 -> 17.32 (gravity only 17.70) | Mixed |
| fine-tune MASt3R on MatrixCity in addition to AerialMegaDepth | MASt3R · MatrixCity (BigCity) | AUC@30: 9.78 -> 29.12 | Improved |
| apply VGA zero-shot on unseen datasets | VGA vs pi3 · ACC-NVS1 and ULTRRA | AUC@30: 43.45 -> 47.45 | Improved |
| apply VGA on very tight rotation/translation thresholds | VGA vs pi3 · ACC-NVS1 and ULTRRA | RTA@2: 1.28 (pi3) vs 0.24 (VGA) | Hurt |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- VGGT — VGGT: Visual Geometry Grounded Transformer (CVPR 2025) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Loc$^{2}$: Interpretable Cross-View Localization via Depth-Lifted Local Feature Matching (ICLR 2026 · similarity 0.89)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.89)
- FG^2: Fine-Grained Cross-View Localization by Fine-Grained Feature Matching (CVPR 2025 · similarity 0.89)
- Seeing Across Skies and Streets: Feedforward 3D Reconstruction from Satellite, Drone, and Ground Images (NeurIPS 2026 · similarity 0.87)
- BevSplat: Resolving Height Ambiguity via Feature-Based Gaussian Primitives for Weakly-Supervised Cross-View Localization (NeurIPS 2025 · similarity 0.87)
- LoD-Loc: Aerial Visual Localization using LoD 3D Map with Neural Wireframe Alignment (NeurIPS 2024 · 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.
