Research map / 3D vision / Depth and stereo
GeoDistill: Geometry-Guided Self-Distillation for Weakly Supervised Cross-View Localization
Shaowen Tong, Zimin Xia, Alexandre Alahi, Xuming He, Yujiao Shi
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 GeoDistill self-distillation | CCVPE · VIGOR Cross-Area | mean localization error (m): 4.97 -> 4.05 | Improved |
| add GeoDistill self-distillation | G2SWeakly (DINO) · VIGOR Cross-Area | mean localization error (m): 3.58 -> 2.68 | Improved |
| add GeoDistill self-distillation | G2SWeakly (VGG) · VIGOR Cross-Area | mean localization error (m): 5.20 -> 4.49 | Improved |
| add GeoDistill self-distillation | CCVPE · KITTI Same-Area | mean localization error (m): 1.28 -> 1.25 | No clear effect |
| add GeoDistill self-distillation | G2SWeakly (VGG) · KITTI Cross-Area | mean localization error (m): 12.54 -> 12.16 | Improved |
| use FoV masking within teacher-student distillation instead of as plain data augmentation (ablation) | G2SWeakly · VIGOR Cross-Area | mean localization error (m): 5.64 -> 4.49 | Improved |
| use FoV masking within teacher-student distillation instead of as plain data augmentation (ablation) | CCVPE · VIGOR Cross-Area | mean localization error (m): 5.37 -> 4.05 | Improved |
| use FoV masking instead of random patch masking (ablation) | GeoDistill (VGG) · VIGOR Cross-Area | mean localization error (m): 5.21 -> 4.49 | Improved |
| use FoV masking instead of random patch masking (ablation) | GeoDistill (DINO) · VIGOR Cross-Area | mean localization error (m): 3.10 -> 2.68 | Improved |
| use FoV masking instead of maximum activation masking (ablation) | GeoDistill (VGG) · VIGOR Cross-Area | mean localization error (m): 5.14 -> 4.49 | Improved |
| add temperature sharpening of heatmaps (ablation) | GeoDistill · VIGOR Cross-Area | mean localization error (m): 5.23 -> 4.49 | Improved |
| distill full heatmap instead of single-mode target (ablation) | GeoDistill · VIGOR Cross-Area | mean localization error (m): 4.96 -> 4.49 | Improved |
Closest papers by meaning
- Cross-View Geo-Localization via Learning Disentangled Geometric Layout Correspondence (AAAI 2023 · similarity 0.89)
- Aligning Geometric Spatial Layout in Cross-View Geo-Localization via Feature Recombination (AAAI 2024 · similarity 0.89)
- Geo2: Geometry-Guided Cross-view Geo-Localization and Image Synthesis (CVPR 2026 · similarity 0.89)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.87)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.86)
- Fully Geometric Panoramic Localization (CVPR 2024 · similarity 0.86)
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.
