Detection and Classification of Asphalt Pavement Deterioration Using YOLOv8

Abstract

Asphalt deterioration is a major problem for road safety and infrastructure maintenance, as it can shorten pavement lifespans, increase driver risks, and cause traffic disruptions. This study applies YOLOv8, a recent object-recognition algorithm, to detect asphalt deterioration across seven classes: Crack, Patch-Crack, Pothole, Patch-Pothole, Net, Patch-Net, and Manhole. Images were processed in grayscale to evaluate performance under low-color-information conditions. Among the YOLOv8 family, the YOLOv8s model performed best, achieving an mAP@50 of 0.963 and an mAP@50-95 of 0.780. A class-wise analysis showed that most classes were detected reliably, with Patch-Crack combining very high recall with the fewest false negatives, whereas the severely under-represented Patch-Net class yielded zero true positives. The results illustrate the usefulness of object-recognition models for infrastructure maintenance, and YOLOv8's balance of speed and accuracy makes it suitable for real-time road-safety and maintenance-planning applications.