A Unified Hierarchical Framework for Fine-grained Cross-view Geo-localization over Large-scale Scenarios

2025-05-12

AI Summary

Existing cross-view geo-localization methods decouple retrieval and metric localization into separate modeling stages, resulting in low task synergy and high training overhead. This paper proposes a unified hierarchical end-to-end framework—the first to jointly optimize both tasks. We design a shared-parameter multi-granularity representation learning mechanism that integrates contrastive and metric learning, and introduce a differentiable loss-guided re-ranking module to strengthen bidirectional task interaction. Evaluated on the VIGOR benchmark, our method achieves substantial gains in 1-meter recall: 39.64% (+38.11%) within-region and 25.58% (+25.15%) cross-region—surpassing all prior state-of-the-art methods. Key contributions include (i) breaking the conventional task-decoupling paradigm, (ii) a novel multi-granularity joint representation architecture, and (iii) a differentiable re-ranking mechanism enabling end-to-end optimization.

Technology Category

• Computer Vision: Image and Video Retrieval
• Machine Learning: Multi-instance/Multi-view Learning
• Search and Optimization: Learning to Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs

Abstract

Cross-view geo-localization is a promising solution for large-scale localization problems, requiring the sequential execution of retrieval and metric localization tasks to achieve fine-grained predictions. However, existing methods typically focus on designing standalone models for these two tasks, resulting in inefficient collaboration and increased training overhead. In this paper, we propose UnifyGeo, a novel unified hierarchical geo-localization framework that integrates retrieval and metric localization tasks into a single network. Specifically, we first employ a unified learning strategy with shared parameters to jointly learn multi-granularity representation, facilitating mutual reinforcement between these two tasks. Subsequently, we design a re-ranking mechanism guided by a dedicated loss function, which enhances geo-localization performance by improving both retrieval accuracy and metric localization references. Extensive experiments demonstrate that UnifyGeo significantly outperforms the state-of-the-arts in both task-isolated and task-associated settings. Remarkably, on the challenging VIGOR benchmark, which supports fine-grained localization evaluation, the 1-meter-level localization recall rate improves from 1.53% to 39.64% and from 0.43% to 25.58% under same-area and cross-area evaluations, respectively. Code will be made publicly available.

Problem

• Unified framework for cross-view geo-localization tasks
• Improves collaboration between retrieval and metric localization
• Enhances fine-grained localization accuracy in large-scale scenarios

Innovation

• Unified hierarchical framework for geo-localization
• Shared parameter learning for multi-granularity representation
• Re-ranking mechanism with dedicated loss function