SeasonStereo: Robust Dense Stereo Matching for Multi-Date Satellite Imagery via Generative AI
2026-07-29 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors created SeasonStereo, a method to make 3D models from satellite images taken at different times and seasons, which usually look quite different. They train their system using fake image pairs that show how appearances change with seasons, instead of needing lots of real, perfectly matched images or detailed ground data. Their approach matches the accuracy of top methods that use expensive LiDAR data, and even produces clearer details. This makes it easier and cheaper to build 3D maps from varied satellite photos over time.
3D reconstructionsatellite imagerystereo matchingdisparity estimationseasonal variationsynthetic training datazero-shot learninggeometric priorsLiDARmulti-date images
Authors
Álvaro Díaz-Laureano, Roger Marí, Elías Masquil, Pablo Arias, Gabriele Facciolo
Abstract
Accurate 3D reconstruction from satellite imagery typically relies on near-simultaneous stereo pairs, limiting its applicability to diachronic settings where multi-date images exhibit varying seasonal and illumination conditions. Training dense stereo matching models robust to appearance changes is a long-standing challenge, as aligned multi-date imagery and ground-truth geometry are costly to obtain at scale. We propose SeasonStereo, a scalable framework that addresses disparity estimation from diachronic satellite images by training on synthetic image pairs with controlled seasonal appearance variation, while leveraging zero-shot geometric priors from foundation models. SeasonStereo matches the accuracy of state-of-the-art LiDAR-supervised models, while producing sharper geometric details without requiring aligned real multi-date training products or LiDAR-derived labels. As a result, SeasonStereo offers a practical path toward large-scale 3D reconstruction from heterogeneous satellite images with reduced supervision cost.