Abstract: Semantic segmentation of remote sensing images aims to assign semantic labels to each pixel in the image, serving as a key technology for achieving precise ground object interpretation.