Detection of Road and Airfield Runway Pathologies Using Artificial Intelligence

Abstract

Road layer quality is critical for transportation efficiency and infrastructure safety. Factors such as temperature fluctuations, rainfall, and the use of inappropriate construction materials can lead to surface deterioration, resulting in cracks and potholes. Traditional inspection methods performed manually by engineers are accurate but costly, labor-intensive, time-consuming, and potentially hazardous. This thesis explores the use of artificial intelligence to achieve automated detection of road defects, focusing on deep learning models, specifically Convolutional Neu- ral Networks (CNNs) and transfer learning, aiming to surpass traditional limitations. The AI-based system utilizes vehicle-mounted cameras, a DVR, and a computer to process video footage for defect detection. Pre-trained CNN models achieve high accuracy, providing a reliable, efficient, and scalable solution for road maintenance. Research indicates that this system outperforms manual methods, enhancing safety and reducing costs. Future work will integrate real-time monitoring and advanced deep learning techniques to improve detection capabilities. This study highlights the significant potential of artificial intelligence in revolutionizing road infrastructure management, pav- ing the way for safe and cost-effective maintenance solutions