Distributed Vehicle Tracking System Using Image Processing Techniques

In this part of the research, an implementation of a distributed vehicle tracking system that uses image processing techniques for real-time automobile surveillance is done. The system uses OpenCV, Pytesseract and Optical Character Recognition (OCR) for the extraction of the license plate information from vehicle images to identify and track vehicles. By doing so, distributed computing enables the system to handle the massive amount of video data on different nodes to increase both efficiency and scalability. The feature of this system is to utilise deep learning models such as YOLO to detect objects, Deep Sort to track vehicles and ReID to recognise the same vehicle across different camera angles. The improved vehicle tracking performance is shown through the distribution of the computing, where the data are processed in real time, and advanced machine learning is applied to ensure reliability and accuracy. It is found that DeepSort is superior to the other models in tracking accuracy, such as occlusion and dense traffic, whereas YOLO excels in the initial detection. The system is evaluated by metric such as rank accuracy, precision, recall, and it is found to be exceedingly reliable in its real-world conditions. Image processing and distributed computing are proven effective for vehicle tracking in research that indicated a promising solution to traffic management and urban safety. In order to improve the future, the constraint of resources and the algorithm to deal with a complicated traffic environment will be taken into consideration.