Traffic Monitoring System Using Drone Images

As a result of the increased density of automobiles with population growth in the modern world, several problems have arisen, including road congestion and air pollution due to emissions from vehicles, along with increased rates of accidents. This is very true for underdeveloped countries where activity monitoring is very difficult. This paper presents a smart traffic monitoring system utilizing some state-of-the-art computer vision methodologies with aerial drone footage. The high-resolution cameras on drones take pictures of the traffic flow, which after processing with YOLOv8 and ResNet50 algorithms for vehicle recognition, showed a 97% and 91% accuracy of identification rate, respectively. Besides, the application of a time series by LSTM and GRUs has performed well in predicting the flow of traffic. Among them, the LSTM model had an MSE of 20.98%, while RMSE accuracy in that produced by GRU was 0.943%. This integrated system will contribute to improving traffic monitoring, congestion identification, and real-time data provision for traffic management. Its future enhancement will enable it to recognize vehicle license plates to facilitate better law and order and surveillance action on the ground.