Global context information is essential for the semantic segmentation of remote sensing (RS) images. However, most existing methods rely on a convolutional neural network (CNN), which is challenging ...