AI-Based System for Real-Time Traffic Flow Assessment and Safety Evaluation on Unsignalized Intersections

Abstract

Real-time traffic monitoring at unsignalized intersections plays an important role in enhancing the performance of traffic management and road safety. However, traditional solutions usually limit themselves only to vehicle detection and counting without adequate assistance in traffic analysis and congestion monitoring. This paper introduces an AI-based framework for real-time traffic flow analysis and safety monitoring based on YOLOv8, OpenCV, and Cellular Automata (CA). The introduced framework makes use of dynamic Region of Interest (ROI) setup with the help of SIDE_LINE and POLYGON modes. Moreover, this framework offers vehicle detection and tracking, key traffic parameters estimation (vehicle count, traffic flow, density, occupancy, queue length), traffic condition classification based on CA congestion analysis model, intelligent safety monitoring, and real-time dashboard with traffic visualization and alert system. The suggested solution was tested with real-world traffic videos obtained from an unsignalized road in Mehsana, Gujarat, India. The results of experimental analysis show that the introduced framework is quite efficient in terms of real-time traffic analysis, congestion estimation, and safety monitoring and can be applied in intelligent transportation systems and smart cities.