<![CDATA[Traffic violations, such as exceeding the speed limit and inappropriate lane usage, are among the main factors causing accidents and congestion on toll roads. To improve traffic safety and efficiency, an automated monitoring system capable of detecting and analyzing violations quickly and accurately is needed. This research aims to develop and evaluate a detection system for speed limit violations and lane misuse by heavy vehicles using deep learning-based object recognition and tracking technology. The method used is the YOLO (You Only Look Once) framework for object detection and the Kalman Filter to track vehicle movement between frames, thereby refining position and speed estimates. The research data was obtained from CCTV video recordings installed along the toll road. The developed system is capable of detecting vehicles, calculating speed based on the shift between frames, and analyzing vehicle position in relation to lane usage regulations. The model evaluation results demonstrated quite good performance with an accuracy of 83.97%, a precision of 0.702, a recall of 0.757, and an F1 score of 0.758. The combination of YOLO and the Kalman Filter proved effective in detecting and tracking vehicles in real time, with adequate accuracy and efficient processing speed. This study concludes that a deep learning-based system can be an innovative solution to support automated traffic monitoring on toll roads. Implementing such a system has the potential to help reduce traffic violations, prevent accidents, and improve driving safety. Furthermore, this study provides recommendations for further development for integration with intelligent transportation technology to support more adaptive and sustainable traffic management. ]]>
