U-Net has been widely applied in semantic segmentation tasks,but it faces challenges in the semantic segmentation of high-resolution remote sensing images due to the loss of boundary information during the downsamplin...U-Net has been widely applied in semantic segmentation tasks,but it faces challenges in the semantic segmentation of high-resolution remote sensing images due to the loss of boundary information during the downsampling process and the inherent blurriness of object boundaries in remote sensing images.We propose an advanced U-Net variant model that addresses these issues.By introducing the CBAM attention mechanism,we enhance the extraction of boundary information during the downsampling process,and by incorporating a cascaded edge detection module,we significantly improve the model’s boundary segmentation performance.As a result,the model demonstrates excellent performance in the segmentation of high-resolution remote sensing images.The results indicate that our proposed model outperforms other baseline models and exhibits superior performance.展开更多
文摘U-Net has been widely applied in semantic segmentation tasks,but it faces challenges in the semantic segmentation of high-resolution remote sensing images due to the loss of boundary information during the downsampling process and the inherent blurriness of object boundaries in remote sensing images.We propose an advanced U-Net variant model that addresses these issues.By introducing the CBAM attention mechanism,we enhance the extraction of boundary information during the downsampling process,and by incorporating a cascaded edge detection module,we significantly improve the model’s boundary segmentation performance.As a result,the model demonstrates excellent performance in the segmentation of high-resolution remote sensing images.The results indicate that our proposed model outperforms other baseline models and exhibits superior performance.
