Development of a YOLOv8-Based Real-Time Vehicle Speed Estimation System for the Universitas Riau Campus

Abstract

This research proposes developing a real-time vehicle speed detection system using the You Only Look Once v8 architecture to monitor speed limit violations on the Universitas Riau campus. The system leverages YOLO's capabilities for real-time detection of fast-moving objects, particularly two- and four-wheeled vehicles, by processing camera video streams on a GPU-based laptop, offering a flexible and cost-effective solution. System testing was conducted at varying vehicle speeds across multiple road locations within the Universitas Riau campus area, demonstrating high detection accuracy with minimal errors and real-time identification of violations. The results indicate an average Mean Absolute Error of 1.2 km/h and Root Mean Square Error of 1.8 km/h for speeds of 30–50 km/h for two- and four-wheeled vehicles under diverse lighting conditions, with detection accuracy reaching 98% and a 100% violation identification success rate, positioning it as an effective solution for campus traffic management to mitigate speeding incidents.