VFM-Loc: Zero-Shot Cross-View Geo-Localization via Aligning Discriminative Visual Hierarchies

TL;DR

VFM-Loc introduces a training-free, zero-shot framework for cross-view geo-localization that aligns discriminative visual features from foundational models, significantly improving accuracy in real-world scenarios with large viewpoint differences.

Contribution

It proposes a novel hierarchical feature extraction and statistical alignment method leveraging vision foundational models for zero-shot cross-view geo-localization.

Findings

Achieves over 20% higher Recall@1 than supervised methods on LO-UCV dataset.

Demonstrates strong zero-shot accuracy on standard benchmarks.

Establishes a training-free, robust paradigm for real-world CVGL.

Abstract

Cross-View Geo-Localization (CVGL) in remote sensing aims to locate a drone-view query by matching it to geo-tagged satellite images. Although supervised methods have achieved strong results on closeset benchmarks, they often fail to generalize to unconstrained, real-world scenarios due to severe viewpoint differences and dataset bias. To overcome these limitations, we present VFM-Loc, a training-free framework for zero-shot CVGL that leverages the generalizable visual representations from vision foundational models (VFMs). VFM-Loc identifies and matches discriminative visual clues across different viewpoints through a progressive alignment strategy. First, we design a hierarchical clue extraction mechanism using Generalized Mean pooling and Scale-Weighted RMAC to preserve distinctive visual clues across scales while maintaining hierarchical confidence. Second, we introduce a statistical manifold alignment pipeline based on domain-wise PCA and Orthogonal Procrustes analysis, linearly aligning heterogeneous feature distributions in a shared metric space. Experiments demonstrate that VFM-Loc exhibits strong zero-shot accuracy on standard benchmarks and surpasses supervised methods by over 20% in Recall@1 on the challenging LO-UCV dataset with large oblique angles. This work highlights that principled alignment of pre-trained features can effectively bridge the cross-view gap, establishing a robust and training-free paradigm for real-world CVGL. The relevant code is made available at: https://github.com/DingLei14/VFM-Loc.