Research map / Detection and segmentation / CNN features
TopicGeo: An Efficient Unified Framework for Geolocation
Xin Wang, Xinlin Wang, Shuiping Gou
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 category embeddings to topic encoder (ablation) | TopicGeo baseline · TZC | R@1: 3.4% improvement | Improved |
| add attribute embeddings (ablation) | TopicGeo with category embeddings · TZC | R@1: adds 2.7% | Improved |
| add category-weighted loss (ablation) | TopicGeo · TZC | R@1: 88.1 -> 90.6 | Improved |
| add LoFTR cross-attention to baseline (ablation) | self-attention baseline · TZC | R@1: 3.7% | Improved |
| add TopicFM self-supervised topics to baseline (ablation) | self-attention baseline · TZC | R@1: 6.2% | Improved |
| use cross-attention encoders instead of lightweight parallel topic branch (ablation) | retrieval encoder · TZC | online retrieval time: 0.87s baseline, 40x slower | Hurt |
| add adaptive center assignment (ablation) | TopicGeo · TZC | initial matching pairs: 10.3% | Improved |
| add MKNN filter (k=1) (ablation) | TopicGeo with ACA · TZC | matching pairs: 12.3% | Improved |
| add global geometric verification (ablation) | TopicGeo · TZC | R@1: 86.5 -> 90.6 | Improved |
| use k=1 MKNN during retrieval (ablation) | TopicGeo with GGV · TZC | R@1: 90.6 -> 87.1 | Hurt |
| add multi-level fine matching (ablation) | TopicGeo · TZC | matching IoU: 85.1 -> 90.5 | Improved |
| embed topics into LoFTR coarse matcher with asymmetric input | LoFTR Coarse · TZC | R@1: 81.6 -> 86.1 | Improved |
Closest papers by meaning
- GeoRanker: Distance-Aware Ranking for Worldwide Image Geolocalization (NeurIPS 2025 · similarity 0.88)
- G3: An Effective and Adaptive Framework for Worldwide Geolocalization Using Large Multi-Modality Models (NeurIPS 2024 · similarity 0.86)
- Vision-Language Reasoning for Geolocalization: A Reinforcement Learning Approach (AAAI 2026 · similarity 0.86)
- Recognition through Reasoning: Reinforcing Image Geo-localization with Large Vision-Language Models (NeurIPS 2025 · similarity 0.86)
- From Coarse to Fine: A Matching and Alignment Framework for Unsupervised Cross-View Geo-Localization (AAAI 2025 · similarity 0.86)
- PIGEON: Predicting Image Geolocations (CVPR 2024 · similarity 0.85)
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.
