Automatic Lane Assistance Line Detection using YOLOX in Vehicle Speed Detection System

Proposedis the optimization of Lane Assistance Linein a vehicle speed detection system based on traffic surveillancevideos using the YOLOX algorithm and automatic homography techniques. The Lane Assistance Line plays a crucial role in calculating vehicle speed as vehicles pass two parallel lines; however, this process is often affected by environmental variables such as video resolution and lighting conditions. By utilizing YOLOX, supporting objects such as road segments, road markings, ground position, and traffic signs can be efficiently detected. The Lane Assistance Line is then projected using automatic homography to improve the accuracy of speed calculations. Experimental results show that YOLOX-l achievesthe highest processing speed (2.16 FPS), while YOLOv9-c offers an optimal balance between speed and accuracy, with an average precision of $74 \%$ and recall of $61 \%$ Despite challenges such as precision in detecting supporting objects and environmental variations, this approach enhances the accuracyof line detection and overall system performance. The YOLOX and automatic homography-based approach makes a significant contribution to the development of a more efficient and accuratevideo-based speed detection system, especially in dynamic trafficconditions.