Analisis Volume Kendaraan Harian Selama SatuMinggu Menggunakan Data Rekaman ATCS Di Kota Palembang

Abstract

The increasing traffic volume in Palembang City, particularly at intersections controlled by the Area Traffic Control System (ATCS), necessitates accurate and efficient traffic analysis to support adaptive signal management. This study aims to analyze daily traffic volume patterns over one week, identify peak hours, and determine vehicle type distribution (cars and motorcycles) using video recordings from ten ATCS points in Palembang. The implemented method is Deep Learning utilizing the YOLOv8 object detection model, combined with Region of Interest (ROI) and Object Tracking techniques for automated vehicle volume counting. The analysis results consistently show that the main traffic peak hour occurs in the afternoon, specifically between 17:00 and 17:10 UTC+7, reflecting commuter mobility patterns. Furthermore, a significant finding is the dominance of two-wheeled vehicles, where the motorcycle volume during peak hours is, on average, twice the volume of cars. The application of the YOLOv8-based method proved effective in providing accurate real-time volume data, making it a crucial input for the Land Transportation Management Agency (BPTD) to optimize signal cycle times and formulate more adaptive transportation policies in the future.