High Security Registration Plate Detection and Classification Using Federated Learning

High Security Registration Plate (HSRP) Classification is essential for enhancing vehicle security, aiding law enforcement, automating toll collection, managing traffic, and ensuring regulatory compliance. Federated Learning enables multiple nodes, each holding localized data, to collaboratively train a global model without sharing raw data. This ensures that privacy is maintained while still benefiting from the collective knowledge of all nodes. Federated Averaging aggregates model updates from each node to create a robust global model. YOLOv8, known for its real-time processing capabilities and high precision, serves as the core detection algorithm, efficiently identifying and classifying HSRPs in images. In this paper HSRP are detected and classified by leveraging the combined strengths of Federated Learning and the YOLOv8 (You Only Look Once, version 8) object detection algorithm. Ensuring data privacy not only protects sensitive information and complies with data protection regulations but also builds trust among users, reduces the risk of data breaches, and encourages collaboration among data custodians. Experimental results indicate that the deployed application using Federated learning and YOLOv8 model achieves a mean Average Precision (mAP) of 99.3%. The system can be enhanced and deployed in traffic monitoring systems and law enforcement applications.