RANCANGAN SISTEM KLASIFIKASI KENDARAAN BERBASIS KAMERA UNTUK ANALISIS DAMPAK LALU LINTAS LINGKUNGAN KAMPUS

Abstract

Traffic congestion in campus environments frequently leads to delays, safety risks, and increased pollution. This study aims to design and test a vehicle classification system based on Power over Ethernet (PoE) cameras utilizing the YOLOv8 algorithm for real-time traffic monitoring in a campus setting. The system integrates edge computing for vehicle detection and classification, coupled with cloud-based dashboard visualization for long-term data analysis. Initial testing was conducted on a residential street as a simulation, using a PoE camera with a resolution of 1920×1080 pixels and a frame rate of 30 fps. The results demonstrated the system’s ability to detect and classify cars and motorcycles with confidence scores ranging from 0.51 to 0.79, identifying a total of 12 vehicles (2 cars and 10 motorcycles) over a 30-minute observation period. The average processing latency of 80 ms supports real-time operation, although challenges such as occlusion and low-speed estimation require further refinement. This study highlights the system’s potential to enhance campus traffic management, with recommendations for future development including model training with local datasets, testing under varied conditions, and integrating predictive analytics.