Road Crack Detection Algorithm Based on Improved YOLOv8

Road crack detection is of great significance for maintaining road quality, improving road safety, optimizing resource allocation and protecting environment. Aiming at the problems of low recognition accuracy, high false detection rate and false detection rate of existing road crack detection, an improved road crack detection algorithm YOLOv8-YP based on YOLOv8 is proposed. Firstly, the attention mechanism GAM is introduced into the backbone network and neck network to enhance the ability of extracting features of different scales from fracture images. Secondly, SPDConv is used to replace Conv in the backbone network to improve the recognition rate of slender and tiny cracks in the backbone network. Finally, DIoU function is used to optimize the bounding box loss of the network, so that the network can converge quickly and provide finer detection results. The experimental results show that the improved algorithm model YOLOv8-YP and the original algorithm YOLOv8n are improved by 2.2% and 2.3% in mAP @ 0.5 and mAP @ 0.5: 0.95, respectively. The improved algorithm YOLOv8-YP has better detection effect on road cracks, and can identify and locate road cracks quickly and accurately.