AI-Driven Computer Vision for Pavement Distress Classification and Structural Degradation Mapping in Urban Highway Networks
Abstract
Timely pavement condition assessment is essential for maintaining road safety and optimizing maintenance planning; however, conventional inspection methods remain labor-intensive, time-consuming, and often lack the capability for continuous large-scale monitoring. This study aims to develop an artificial intelligence-based pavement monitoring framework that integrates real-time pavement distress detection, Pavement Condition Index (PCI) estimation, and Geographic Information System (GIS)-based spatial visualization into a unified decision-support platform. The proposed framework employs a YOLO-based deep learning model for automated detection of multiple pavement distress types using vehicle-mounted camera imagery, followed by PCI computation and geospatial degradation mapping to support network-level pavement assessment. Experimental evaluation demonstrated that the proposed model achieved a composite precision of 0.923, recall of 0.881, mean Average Precision (mAP) of 0.769, and an average inference time of 8.4 ms, indicating its capability to perform accurate real-time pavement distress detection under diverse urban road conditions. The novelty of this research lies in the integration of AI-based pavement distress detection, automated PCI estimation, and GIS-based structural degradation mapping within a single low-cost vehicle-mounted inspection framework, enabling simultaneous condition assessment and spatial infrastructure management without relying on expensive inspection technologies. This integrated workflow provides a practical contribution by supporting maintenance prioritization, reducing inspection time, and improving data-driven pavement management for transportation agencies. The results demonstrate that the proposed framework offers an effective and scalable solution for intelligent pavement monitoring while extending the application of computer vision beyond object detection toward comprehensive infrastructure condition assessment and maintenance decision support.