SU²-Net: Siamese-Based U²-Net for Change Detection in Remote Sensing Images

Change detection in remote sensing images is an important research topic for many fields, such as urban planning and disaster management, as it enables environmental monitoring. The increase in the performance of deep learning methods has brought about advances in this field. In this study, the Siamese-based U 2 -Net (SU 2 -Net) model is proposed for change detection in high-resolution remote sensing images. In order to successfully detect changes without significantly increasing the computational cost with a deep network model, the existing U 2 -Net model was transformed into a Siamese-based architecture. Change detection dataset (CDD) was used in the study. In order to prevent the unbalanced data set problem, changes were detected by using the hybrid loss function. As a result of the experiments, 0.953 precision, 0.946 recall, and 0.949 F1-score values were obtained with the SU 2 -Net model. Compared to existing methods, our proposed model has been shown to have high performance with improvements in computational cost and an F1-score value of 0.949.