Seg-Road: A Segmentation Network for Road Extraction Based on Transformer and CNN with Connectivity Structures

Acquiring road information is important for smart cities and sustainable urban development. In recent years, significant progress has been made in the extraction of from remote sensing images using deep learning (DL) algorithms. However, due to complex shape, narrowness, high span roads images, results are often unsatisfactory. This article proposes a Seg-Road model improve connectivity. The uses transformer structure extract long-range dependency global contextual fragmentation segmentation convolutional neural network (CNN) local details. Furthermore, novel pixel connectivity (PCS) proposed robustness prediction results. To verify effectiveness segmentation, DeepGlobe Massachusetts datasets were used training testing. experimental show that achieves state-of-the-art (SOTA) performance, with an intersection over union (IoU) 67.20%, mean (MIoU) 82.06%, F1 91.43%, precision 90.05%, recall 92.85% dataset, IoU 68.38%, MIoU 83.89%, 90.01%, 87.34%, 92.86% which better than values CoANet. Further, it higher application value achieving