Application of YOLO Model in Road Defect Detection

Identifying road defects quickly, easily and efficiently is of great significance to ensuring road safety and avoiding accidental casualties and property losses. Compared with other methods of detecting road defects through human eyes or photo recognition, YOLO has the advantages of real-time performance, high accuracy, simultaneous multi-target detection, and simple and efficient models. This article introduces several models of YOLOv5s, YOLOv5m, YOLOv5l, YOLOv5x, and YOLOv7, which are combined with drones to capture images and integrate and analyze road crack and bump data. And tested the analysis speed of YOLOv7 for test images. The results show that the YOLOv7 model is the most dominant.