OffNadirLoc: Benchmark and Framework for Challenging UAV-to-Satellite Geo-Localization under Large Off-Nadir Views

2026-07-22

AI Summary

This work addresses the challenge of cross-view geolocalization under large off-nadir angles, where severe viewpoint distortion, occlusion, and appearance discrepancies hinder matching between drone and satellite imagery. To this end, we introduce OffNadirLoc, the first benchmark specifically designed for such scenarios, along with a novel framework termed ONLoc. ONLoc incorporates a structure-aware context weighting mechanism and a view-consistent learning strategy, enhanced by multi-view semantic group supervision to overcome the limitations of conventional pairwise contrastive learning. This enables the learning of viewpoint-invariant yet discriminative feature representations. Extensive experiments demonstrate that ONLoc consistently outperforms existing methods on both OffNadirLoc and four near-nadir datasets, while also exhibiting strong zero-shot transfer capability—generalizing effectively to unseen datasets without fine-tuning.

Abstract

Cross-view geo-localization between UAV and satellite imagery remains a fundamental yet highly challenging task, especially under large off-nadir views where drastic perspective distortions, occlusions, and appearance gaps occur. Existing benchmarks and methods primarily focus on near-nadir scenarios and often overlook the importance of structural scene understanding and intra-domain relational constraints, limiting their performance in real-world deployments. In this work, we introduce OffNadirLoc, a new benchmark for large off-nadir UAV-to-satellite geo-localization. To tackle the unique challenges posed by off-nadir perspectives, we further propose ONLoc, a framework that incorporates a structure-aware contextual weighting mechanism to dynamically emphasize reliable local features while suppressing ambiguous or repetitive regions. Additionally, we design a view-coherent learning strategy, which treats one satellite image and the corresponding UAV images from multiple views as a cohesive semantic group. This set-level supervision enables the model to learn viewpoint-invariant and discriminative features, making it more effective at capturing multi-view consistency than conventional pairwise contrastive learning. Extensive experiments on the OffNadirLoc benchmark and four near-nadir datasets demonstrate that our method consistently outperforms state-of-the-art approaches while exhibiting strong zero-shot generalization to unseen datasets without additional training. The code will be released at https://montalario.github.io/offnadirloc/.

Problem

• UAV-to-satellite geo-localization
• off-nadir views
• cross-view geo-localization
• perspective distortion
• appearance gap

Innovation

• off-nadir geo-localization
• structure-aware weighting
• view-coherent learning
• cross-view retrieval
• zero-shot generalization