Cross-View Image Set Geo-Localization

TL;DR

This paper introduces a new cross-view image set geo-localization task, a large benchmark dataset, and a flexible method that improves localization accuracy by integrating multiple perspectives and geo-attributes.

Contribution

The paper proposes the novel Set-CVGL task, creates the extensive SetVL-480K benchmark, and introduces FlexGeo, a versatile method that enhances geo-localization by fusing multi-view features and leveraging geo-attributes.

Findings

FlexGeo outperforms existing methods on multiple datasets.

Localization accuracy improves by over 22% on SetVL-480K.

The method effectively integrates diverse perspectives for better localization.

Abstract

Cross-view geo-localization (CVGL) has been widely applied in fields such as robotic navigation and augmented reality. Existing approaches primarily use single images or fixed-view image sequences as queries, which limits perspective diversity. In contrast, when humans determine their location visually, they typically move around to gather multiple perspectives. This behavior suggests that integrating diverse visual cues can improve geo-localization reliability. Therefore, we propose a novel task: Cross-View Image Set Geo-Localization (Set-CVGL), which gathers multiple images with diverse perspectives as a query set for localization. To support this task, we introduce SetVL-480K, a benchmark comprising 480,000 ground images captured worldwide and their corresponding satellite images, with each satellite image corresponds to an average of 40 ground images from varied perspectives and locations. Furthermore, we propose FlexGeo, a flexible method designed for Set-CVGL that can also adapt to single-image and image-sequence inputs. FlexGeo includes two key modules: the Similarity-guided Feature Fuser (SFF), which adaptively fuses image features without prior content dependency, and the Individual-level Attributes Learner (IAL), leveraging geo-attributes of each image for comprehensive scene perception. FlexGeo consistently outperforms existing methods on SetVL-480K and two public datasets, SeqGeo and KITTI-CVL, achieving a localization accuracy improvement of over 22% on SetVL-480K.