Road Damage Detection using YOLO

Abstract

This study presents a novel approach for the automatic detection of road surface damage using UAV imagery combined with deep learning techniques. Maintaining road infrastructure is essential for ensuring safe and efficient transportation; however, traditional inspection methods are labor-intensive, slow, and sometimes dangerous for personnel. To overcome these challenges, the proposed framework integrates unmanned aerial vehicles (UAVs) with artificial intelligence to improve both detection speed and accuracy. The method applies advanced object detection models, including YOLOv5, YOLOv7, and the improved YOLOv8 algorithms, to analyze UAV-captured road images. These models were trained and evaluated using publicly available datasets collected from road environments in China and Spain. Experimental results demonstrate strong performance, achieving detection accuracy levels of up to 85%. The findings highlight the effectiveness of combining UAV technology with deep learning methods for automated road damage identification and emphasize its potential to support smarter infrastructure monitoring systems and future research developments.