Abstract: Deep learning methods, especially convolutional neural networks (CNNs) and vision transformers (ViTs), are frequently employed to perform semantic segmentation of high-resolution remotely ...