UAVSat-Deg: Robustness Benchmark for Degraded UAV-Satellite Geo-Localization

Researchers introduced UAVSat-Deg, a large-scale robustness benchmark for UAV-satellite cross-view geo-localization under real-world image degradations. The benchmark includes two datasets, University-1652-Deg and SUES-200-Deg, covering 27 corruption types (19 core and 8 compound) at three severity levels. It supports bidirectional drone-to-satellite and satellite-to-drone retrieval and multi-height UAV acquisition, containing over 11.7 million pre-generated corrupted test images. The work highlights that existing methods achieve high accuracy on clean benchmarks but remain largely unexamined under adverse weather, illumination changes, platform motion, sensor noise, and compression. The benchmark aims to evaluate and improve robustness of geo-localization models in real-world flight conditions.