Purpose: This study aims to develop a traffic density prediction system in Bandar Lampung City to address increasing congestion caused by the rapid growth of vehicles that exceeds road capacity. The system is intended to support real-time monitoring, improve traffic management efficiency, and facilitate data-driven decision-making for adaptive traffic light control and route diversion.

Research Methodology: The study employed an experimental approach combined with prototyping. Vehicle detection was performed using the YOLO algorithm on CCTV footage collected from congestion-prone areas. The resulting data were processed and visualized through a web-based dashboard. System performance was evaluated based on vehicle detection accuracy and real-time processing speed under various traffic conditions.

Results: The developed system successfully detected vehicles from CCTV footage in real-time and displayed traffic density information through an interactive web dashboard. The system enabled adaptive traffic management by providing authorities with accurate and timely data on congestion patterns.

Conclusions: The study demonstrates that integrating YOLO-based vehicle detection with a web-based dashboard improves traffic management efficiency in Bandar Lampung City. Real-time monitoring and data visualization enhance the ability of authorities to make informed, timely decisions, contributing to more effective traffic control.

Limitations: The study is limited by the use of CCTV footage from selected congestion-prone areas, a relatively small dataset, and potential variability in detection accuracy under extreme weather or low-light conditions.

Contribution: This research provides a practical model for real-time traffic monitoring and management using YOLO and web-based visualization. The system offers a replicable framework for other urban areas facing similar traffic congestion challenges and supports data-driven policymaking.