AG-SemiCD: A Semi-Supervised Remote Sensing Change Detection Framework Integrating Adaptive Confidence Filtering and Attention-Guided Feature Aggregation (opens in new tab)

Supervised change detection methods excel at extracting change features from remote sensing. However, their heavy reliance on large-scale, high-quality labeled data incurs high costs and time consumption, thereby restricting their practical application and scalability. To fully leverage limited labeled data and abundant unlabeled data, this article proposes AG-SemiCD, a semi-supervised change detection framework for remote sensing that combines adaptive confidence filtering with attention-gui...