Research map / Label-efficient and robust learning / Self-supervised and contrastive
Viewpoint Invariant Dense Matching for Visual Geolocalization
Gabriele Berton, Carlo Masone, Valerio Paolicelli, Barbara Caputo
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 GeoWarp re-ranking to GeM | AlexNet GeM · R-Tokyo 25m | recall@1: 23.2 -> 34.0 | Improved |
| add GeoWarp re-ranking to NetVLAD | ResNet-50 NetVLAD · R-Tokyo 25m | recall@1: 72.4 -> 74.4 | Improved |
| add GeoWarp re-ranking to NetVLAD | VGG16 NetVLAD · Pitts30k 25m | recall@1: 82.5 -> 83.3 | Improved |
| add self-supervised loss to warping training (ablation) | ResNet50 NetVLAD GeoWarp · Pitts30k val | recall@1: 80.7 -> 87.0 | Improved |
| re-rank with local features instead of global features, untrained (ablation) | ResNet50 NetVLAD · Pitts30k val | recall@1: 86.4 -> 80.7 | Hurt |
| add consistency loss alone, without self-supervision (ablation) | VGG16 GeM GeoWarp · Pitts30k val | recall@1: 74.3 -> 74.2 | No clear effect |
| add feature-weighted weak loss alone, without self-supervision (ablation) | VGG16 GeM GeoWarp · Pitts30k val | recall@1: 74.3 -> 74.3 | No clear effect |
| add both weak losses on top of self-supervised loss (ablation) | VGG16 GeM GeoWarp · Pitts30k val | recall@1: 78.6 -> 79.0 | Improved |
| add QE + DBA query expansion to NetVLAD | ResNet-50 NetVLAD · Pitts30k 25m | recall@1: 84.3 -> 83.7 | Hurt |
| add InLoc re-ranking instead of NetVLAD alone | VGG16 NetVLAD · Pitts30k 25m | recall@1: 82.5 -> 72.8 | Hurt |
| use Rocco et al. warping instead of GeoWarp warping module | VGG16 NetVLAD · Pitts30k 25m | recall@1: 83.3 -> 81.9 | Hurt |
| increase warping coefficient k (ablation) | GeoWarp · coarser thresholds | recall@1 | Mixed |
Closest papers by meaning
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.86)
- OpenStreetView-5M: The Many Roads to Global Visual Geolocation (CVPR 2024 · similarity 0.86)
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.85)
- Around the World in 80 Timesteps: A Generative Approach to Global Visual Geolocation (CVPR 2025 · similarity 0.85)
- G3: An Effective and Adaptive Framework for Worldwide Geolocalization Using Large Multi-Modality Models (NeurIPS 2024 · similarity 0.84)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.84)
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.
