CLNet: Cross-View Correspondence Makes a Stronger Geo-Localizationer

2025-12-16

AI Summary

Cross-view geolocalization (IRCVGL) faces significant challenges due to large viewpoint disparities between satellite and street-level imagery, hindering explicit spatial correspondence modeling. To address this, we propose a correspondence-aware feature refinement framework that introduces a novel neural correspondence map to model pixel-level spatial mappings. This is integrated with a nonlinear embedding transformer for semantically consistent cross-view feature alignment and a global feature recalibration mechanism to enhance geometric consistency. Our method jointly leverages implicit correspondence field modeling, MLP-driven feature transformation, and spatially guided channel reweighting. Evaluated on four major benchmarks—CVUSA, CVACT, VIGOR, and University-1652—it achieves state-of-the-art performance, substantially improving localization accuracy, generalization capability, and interpretability.

Technology Category

• Computer Vision: Multi-modal Vision
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Localization, Mapping, and Navigation

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs

Abstract

Image retrieval-based cross-view geo-localization (IRCVGL) aims to match images captured from significantly different viewpoints, such as satellite and street-level images. Existing methods predominantly rely on learning robust global representations or implicit feature alignment, which often fail to model explicit spatial correspondences crucial for accurate localization. In this work, we propose a novel correspondence-aware feature refinement framework, termed CLNet, that explicitly bridges the semantic and geometric gaps between different views. CLNet decomposes the view alignment process into three learnable and complementary modules: a Neural Correspondence Map (NCM) that spatially aligns cross-view features via latent correspondence fields; a Nonlinear Embedding Converter (NEC) that remaps features across perspectives using an MLP-based transformation; and a Global Feature Recalibration (GFR) module that reweights informative feature channels guided by learned spatial cues. The proposed CLNet can jointly capture both high-level semantics and fine-grained alignments. Extensive experiments on four public benchmarks, CVUSA, CVACT, VIGOR, and University-1652, demonstrate that our proposed CLNet achieves state-of-the-art performance while offering better interpretability and generalizability.

Problem

• Explicitly models spatial correspondences for cross-view geo-localization
• Bridges semantic and geometric gaps between satellite and street-level images
• Enhances interpretability and generalizability in image retrieval-based localization

Innovation

• Neural Correspondence Map aligns cross-view spatial features
• Nonlinear Embedding Converter remaps features across perspectives
• Global Feature Recalibration reweights channels using spatial cues