In today's setting, it is crucial to address concerns and tackle the negative impacts of traffic congestion and inefficiency. Smart cities aim to achieve transportation and reduce traffic congestion, which requires traffic monitoring systems. These systems play a role, in providing real-time data for decision-making. This research paper introduces a “Traffic Monitoring System for Smart Cities” that is based on the concept of estimating traffic density. By utilizing cutting-edge machine learning models like YOLOv8 and ByteTrack this system has been trained extensively over 100 epochs to accurately estimate real-time traffic data. Visualizing the data through heat maps provides insights into traffic density patterns, throughout the city. The results are evaluated on the newly introduced India driving dataset (IDD). The efficiency of the model is effectively shown by the normalized confusion matrix and precision-recall confidence curve. It provides an accuracy of more than 80% for frequently seen vehicles on challenging datasets of real-time driving scenarios on Indian roads with variable backgrounds.