UAV Based Real-Time Implementation of Vehicle Trajectory Tracking Systems

In India, the persistent challenges in traffic management, characterized by congestion, erratic driving behaviour, and frequent violations of traffic regulations, underscore the urgent need for advanced solutions to enhance road safety and efficiency. To address this problem, this project focuses on leveraging deep learning techniques and UAV, to predict the potential likelihood of vehicle collisions based on their predicted future actions. This research involves analyzing a dataset sourced from Ho Chi Minh City (HCMC). The primary goal is to develop a system capable of predicting specific vehicle trajectories utilizing YoloV8, DeepSort and LSTM architectures. This system achieves an accuracy of 99.38%. This automated system can be implemented with a defined set of traffic rules and regulations allowing for automated fines for non-compliant behavior.