STEAM: Stable Self-Training with Elastic Matching and Adaptive Purification
2026-07-10 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors developed a new method called STEAM to help drones figure out their location by matching their view of the ground with satellite images, without needing GPS. Unlike other methods that require lots of labeled image pairs or complicated guesses, STEAM learns directly from real images using a special self-training process. It uses smart techniques to keep the image features stable, find good matching pairs, and clean up mistakes as it learns. Their tests show this method works better than other unsupervised methods and almost as well as those needing labeled data.
cross-view geo-localizationdrone-view imagessatellite-view imagesself-trainingpseudo-labelsfeature representationunsupervised learningelastic matchingadaptive purificationcomputer vision
Authors
Shaoxiang Wang, Kejia Zhang, Haiwei Pan, Lan Zhang
Abstract
Cross-view geo-localization (CVGL) aims to achieve GPS-free localization by matching drone-view images with corresponding satellite-view images. Existing supervised methods rely on large-scale manually annotated cross-view image pairs, making them costly and difficult to scale. In contrast, existing unsupervised approaches typically depend on generative models or clustering-based stage-wise optimization, which are prone to distribution bias and the accumulation of noisy pseudo-labels. To address these limitations, we propose STEAM (Stable Self-Training with Elastic Matching and Adaptive Purification), an end-to-end unsupervised cross-view geo-localization framework that performs self-training directly on real drone and satellite images. Specifically, the proposed Stable Spatial-Aware Module enhances the stability of feature representations, Elastic Matching discovers high-quality cross-view pseudo-labels, and Adaptive Purification dynamically maintains a reliable pseudo-label repository throughout the self-training process. Extensive experiments on the University-1652 and SUES-200 benchmarks demonstrate that STEAM achieves state-of-the-art performance among all existing unsupervised methods and delivers performance comparable to supervised approaches, validating the effectiveness and superiority of the proposed framework. The source code is available at https://github.com/wsx-heu/STEAM.git.