A Real-Time Framework for Traffic Control in Resource-Constrained Cities

Abstract

Rapid urbanization in low-and middle-income countries has intensified traffic congestion and exposed the limitations of traditional traffic management systems, which often rely on static signal timing and manual monitoring. To address these challenges, this paper presents a real-time, end-to-end framework for intelligent traffic control in resource-constrained urban environments. The system integrates computer vision, fuzzy logic, and edge computing to enable adaptive signal regulation and continuous monitoring. It operates in two distinct phases: (1) offline model training, which involves preprocessing and learning from benchmark datasets such as KITTI, UA-DETRAC, and historical traffic records. This phase includes training YOLOv8 for object detection, DeepSORT for tracking, LSTM for temporal prediction, and Random Forest for classification, followed by validation and optimization to produce deployment-ready models. (2) online real-time inference, where live inputs including video streams, sensor data, and traffic conditions are processed through the pre-trained models in a sequential pipeline that culminates in fuzzy logic-based control decisions. The system is implemented using a Flask-based REST API backend and a React.js frontend, supporting scalable deployment and intuitive operator interfaces. Experimental validation using simulated urban traffic videos and benchmark datasets demonstrated real-time performance at 25-40 frames per second with latency ranging from 20-35 ms per frame. Detection precision exceeded 90%, recall reached approximately 88%, and tracking success was around 95% with less than 5% error. Adaptive control maintained a flow rate of 32.5 vehicles per hour and a density of 80.0 vehicles per kilometer during congestion scenarios. This work offers a robust, low-cost, and scalable solution for intelligent traffic management, demonstrating the full lifecycle from offline model preparation to real-time deployment, and supporting sustainable mobility and local capacity building in developing urban contexts.