Performance review of RTI IMS software for automatic road surface damages identification

Abstract

Nowadays, the rapid development of AI models (Artificial Intelligence) has significantly improved performance in addressing construction management challenges in daily life. Among the construction management tasks, the detection of road surface damage in the management of the transportation system has become more straightforward due to robust advancements in machine learning (ML) identification techniques. In this study, the innovative YOLO technology is utilized to create the RTI IMS software, capable of automatically and accurately detecting road surface damages. The research also used the Mosaic method to construct an image library of road surface damages, tripling the variations, markedly enhancing prediction accuracy and improving the software’s detection performance. To illustrate the software’s superiority, the study evaluates object detection performance and speed compared to previous models for detecting road damage. The research culminates in the successful development of comprehensive software capable of automatic, accurate and rapid detection of road surface damages. The RTI IMS software can identify road surface damages and provide defects images to traffic management agencies, contributing to the increased efficiency of these agencies in patrolling, detecting and maintaining surface road defects.