Automated Traffic Control for Sustainable Urban Mobility

Nowadays, urban areas worldwide face several significant challenges in managing traffic, which leads to increases fuel consumption, pollution and travel time. Video-based object detection can collect useful data from video frames like vehicle count, speed, type of vehicle, license plates and more. This can be done efficiently and in a cost-effective way and will help improve safety and traffic management. Traditional systems rely on multiple older algorithms and different sensors to capture data that lacks accuracy. The proposed system integrates computer vision techniques and deep learning models like YOLOv8 using a single Convolutional Neural Network that divides the image into a grid. The count of the detected objects is further stored and using K-Means Clustering Algorithm, data is analyzed and further used for alerting the users about places with heavy congestion. Based on this congestion data, traffic lights will be changed based on the severity of the congestion instead of using regular timer-based traffic signals, thus, improving traffic control and safety.