Barlow twin self-supervised pre-training for remote sensing change detection

Remote sensing change detection (CD) methods that rely on supervised deep convolutional neural networks require large-scale labelled data, which is time-consuming and laborious to collect and label, especially for bi-temporal samples containing changed areas. Conversely, acquiring a large volume of unannotated images is relatively easy. Recently, self-supervised contrastive learning has emerged as a promising method for learning from unannotated images, thereby reducing the need for annotation. However, most existing methods employ random values or ImageNet pre-trained models to initialize their encoders and lack prior knowledge tailored to the demands of CD tasks, thus constraining the performance of CD models. To address these challenges, we propose a novel Barlow Twins self-supervised pre-training method for CD (BTSCD), which uses absolute feature differences to directly learn distinct representations associated with changed regions from unlabelled bi-temporal remote sensing images in a self-supervised manner. Experimental results obtained using two publicly available CD datasets demonstrate that our proposed approach exhibits competitive quantitative performance. Moreover, the proposed method achieved final results superior to those of existing state-of-the-art methods.