Intelligent Application for Road Damage Detection using Deep Learning

Road infrastructure is vital in ensuring transportation efficiency and public safety. Timely detection and maintenance of road damages such as cracks, potholes, and surface wear are essential to prevent accidents and reduce vehicle maintenance costs. This paper presents an intelligent application for automated road damage detection leveraging deep learning techniques. In this paper, the proposed method, YOLOv8, is used for road damage detection, and a convolutional neural network is used for road quality classification. Image denoising is achieved by a denoising autoencoder. The experimental results demonstrate that road damage detection using YOLOv8 shows a 91% confidence score on D40-Rutting, bump, pothole, and separation types of damage and road quality classification; CNN models show 91.12% accuracy. This research contributes toward developing scalable and cost-effective solutions for intelligent transportation systems and smart city infrastructure maintenance.