AI-Powered
Real-Time Traffic Monitoring & Prediction Solution for Global Smart Cities

This project is a case study of building an intelligent vehicle detection and traffic volume prediction system utilizing local CCTV infrastructure to resolve traffic congestion in major urban areas of Malaysia and maximize road operational efficiency.

TOSKY optimized an advanced YOLO-based deep learning model for real-time identification of various vehicle classes suited to Malaysian road characteristics, including passenger cars, trucks, and motorcycles.

Advanced data preprocessing and training processes were conducted considering challenging environmental variables such as drastic illumination changes between day and night, tropical weather conditions, and various camera angles, achieving detection performance (AP) above 90%.
Beyond simply detecting vehicles, the collected data was extended into a model that predicts traffic congestion by time period through time-series analysis techniques.
This completed an integrated monitoring environment including an intuitive visualization dashboard, enabling traffic authorities to establish scientific, data-driven road policies and manage infrastructure efficiently.