<![CDATA[Objective: To develop an artificial intelligence-based detection system for identifying Over-Dimension and Overload (ODOL) vehicles using the YOLO algorithm. Specifically, this study focused on improving the efficiency of ODOL vehicle monitoring, which is important for maintaining road safety, reducing infrastructure damage, and supporting effective transportation regulation. This research contributes to the development of smart transportation systems and sustainable infrastructure innovation in accordance with Sustainable Development Goal 9 (Industry, Innovation, and Infrastructure). Method: The study employed a computer vision-based approach using YOLOv8m and YOLOv10n algorithms for ODOL vehicle detection. The dataset consisted of vehicle images containing over-dimension vehicles, normal vehicles, and trucks collected from digital image sources. Results: The experimental results showed that YOLOv8m achieved better performance compared with YOLOv10n in detecting ODOL vehicles. YOLOv8m obtained a confusion matrix value of 78%, a precision-recall curve value of 81.7%, precision of 91.6%, and recall of 90%. Although YOLOv10n achieved a higher recall value of 93%, its overall detection performance was lower, with a confusion matrix value of 59%, precision-recall curve value of 72.3%, and precision of 89.9%. The implementation of YOLOv8m into an IoT-based detection system successfully enabled real-time data transmission and storage in a database, demonstrating its capability for practical ODOL vehicle monitoring applications. Novelty: The study provides a novel implementation of YOLO-based artificial intelligence technology for real-time ODOL vehicle detection by integrating deep learning, computer vision, and IoT infrastructure. The developed system supports the advancement of intelligent transportation infrastructure and contributes to sustainable innovation in transportation management in line with SDG 9 (Industry, Innovation, and Infrastructure).]]>
