Research map / 3D vision / Depth and stereo
Uncertainty-Aware Vision-Based Metric Cross-View Geolocalization
Florian Fervers, Sebastian Bullinger, Christoph Bodensteiner, Michael Arens, Rainer Stiefelhagen
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 pseudo-label ground truth poses (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 2.37 -> 1.19 | Improved |
| add BEV upsampling (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 2.63 -> 1.19 | Improved |
| add encoder pretraining (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 4.15 -> 1.19 | Improved |
| add vehicle images (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 11.95 -> 1.19 | Improved |
| add self-attention block (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.38 -> 1.19 | Improved |
| add deformable offsets (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.22 -> 1.19 | Improved |
| add data pruning (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.23 -> 1.19 | Improved |
| use 4 heads instead of 1 (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.23 -> 1.19 | Improved |
| use 3 refinement blocks instead of 1 (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.24 -> 1.19 | Improved |
| direct attention logits instead of QKV attention (ablation) | Full model · Palo Alto/San Francisco | mean error (m): 1.25 -> 1.19 | Improved |
| train cross-area and cross-vehicle instead of same-area | Ours · Ford AV Log1 | lateral recall@1m (%): 96.3 -> 77.0 | Hurt |
| remove vehicle camera input (prior only) (ablation) | Ours · Ford AV Log1 | lateral recall@1m (%): 96.3 -> 15.1 | Hurt |
Closest papers by meaning
- Beyond Cross-View Image Retrieval: Highly Accurate Vehicle Localization Using Satellite Image (CVPR 2022 · similarity 0.90)
- UniGeoRS: A Unified Benchmark for Tri-view Geo-Localization (CVPR 2026 · similarity 0.89)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.89)
- PIDLoc: Cross-View Pose Optimization Network Inspired by PID Controllers (CVPR 2025 · similarity 0.87)
- CVGL: Causal Learning and Geometric Topology (NeurIPS 2025 · similarity 0.87)
- Weather-Robust Cross-View Geo-Localization via Prototype-Based Semantic Part Discovery (NeurIPS 2026 · 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.
