Research map / Detection and segmentation / Open-set and few-shot detection
InfoGeo: Information-Theoretic Object-Centric Learning for Cross-View Generalizable UAV Geo-Localization
Hongyang ZHANG, Maonan Wang, Ziyao Wang, Hongrui Yin, Man On Pun
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 OCVA object-centric visual augmentation (ablation) | baseline (same params) · SUES-200 150 m (University-1652 -> SUES) | R@1: 86.10 -> 87.80 | Improved |
| add L_cacs concept selection loss (ablation) | OCVA-only · SUES-200 150 m (University-1652 -> SUES) | R@1: 87.80 -> 88.82 | Improved |
| add L_struct (CSRR) (ablation) | OCVA + L_cacs · SUES-200 150 m (University-1652 -> SUES) | R@1: 88.82 -> 90.87 | Improved |
| add relational distillation (RD) (ablation) | InfoGeo · SUES-200 150 m (University-1652 -> SUES) | R@1: 90.87 -> 91.80 | Improved |
| add relational distillation (RD) (ablation) | InfoGeo · DenseUAV -> SUES-200 (150 m) | R@1: 88.20 -> 88.70 | Improved |
| replace L_struct with InfoNCE for concept alignment (ablation) | CSRR · SUES-200 150 m (University-1652 -> SUES) | R@1: 91.80 -> 88.20 | Hurt |
| use L_struct instead of no CSRR (ablation) | InfoGeo · SUES-200 150 m (University-1652 -> SUES) | R@1: 88.65 -> 91.80 | Improved |
| raise spectral dimension r from 4 to 8 (ablation) | L_struct · SUES-200 150 m (University-1652 -> SUES) | R@1: 91.80 -> 86.42 | Hurt |
| lower spectral dimension r from 4 to 2 (ablation) | L_struct · SUES-200 150 m (University-1652 -> SUES) | R@1: 91.80 -> 89.22 | Hurt |
| increase slot number K to 32 (ablation) | OCVA · SUES-200 150 m | R@1 | Hurt |
| set alpha to 0.8 versus intermediate values (ablation) | OCVA · SUES-200 150 m | R@1 | Improved |
| use RD at inference instead of slot modules | InfoGeo · Computational cost | Params (M): 121.01 -> 90.50 | Improved |
Closest papers by meaning
- Weather-Robust Cross-View Geo-Localization via Prototype-Based Semantic Part Discovery (NeurIPS 2026 · similarity 0.89)
- PAUL: Uncertainty-Guided Partition and Augmentation for Robust Cross-View Geo-Localization under Noisy Correspondence (CVPR 2026 · similarity 0.89)
- Game4Loc: A UAV Geo-Localization Benchmark from Game Data (AAAI 2025 · similarity 0.88)
- MMGeo: Multimodal Compositional Geo-Localization for UAVs (ICCV 2025 · similarity 0.88)
- View From Above: Orthogonal-View aware Cross-view Localization (CVPR 2024 · similarity 0.87)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 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.
