Multi-Grade Road Distress Detection Strategy Based on Enhanced YOLOv8 Model

The total mileage of the road network in China has been growing rapidly during the last twenty years. With the development of deep learning, the automatic road distr ess detection method is more accurate and effective than manual detection. However, the road are classified into five grades according to the Chinese road standard and each grade has its own characteristics. A single model cannot effectively identify multi-grade roads with different materials and levels of road distress. This study proposes a YOLOv8-based road distress detection strategy adapted for multiple road grades. The improved URetinex-Net network is used to enhance the spatial features and scenario diversity of the road distress datasets. Compared to the base YOLOv8 model, the enhancements have led to a 12% increase in accuracy for cement roads, a 22.3% improvement in detection speed, a 5.5% increase in accuracy for ordinary asphalt roads, a 7.5% increase in recognition accuracy for highways, and a 9.3% improvement in detection speed, with significant effects. This study refines the classification of roads based on their grades and matches them with corresponding artificial intelligence training strategies, providing guidance for road inspection and maintenance.

Dieses Werk wurde unter der Creative-Commons-Lizenz Namensnennung 4.0 International (CC-BY 4.0) veröffentlicht und darf unter den Lizenzbedinungen vervielfältigt, verbreitet, öffentlich zugänglich gemacht, sowie abgewandelt und bearbeitet werden. Dabei muss der Urheber bzw. Rechteinhaber genannt und die Lizenzbedingungen eingehalten werden.