Automatic Number Plate and Face Recognition System for Secure Gate Entry into Military Establishments

There is a dire need for an automated, real-time mechanism for ensuring secure, authenticated gate entry of authorized personnel into high-security premises. Identification of the person as well as their vehicle is important for the security requirements of these premises. Considering these needs, we propose an integrated real-time system that employs both an Automatic Number Plate Recognition (ANPR) system and a Face Recognition system for secure gate entry. The designed YOLOv8 model for the ANPR system is trained on Indian number plate datasets assembled by gathering standard open datasets available from public resources. This improves the generalization capability of AI models to detect and recognize number plates accurately. Acknowledging the prevalence of nonstandard number plates in India, a custom dataset is curated and annotated. To avoid the risk of bypassing a single authentication mechanism, two-factor authentication using an additional security level based on Face Recognition is proposed. It uses the dlib face recognition CNN model to verify the identity of individuals. The proposed system registers vehicle and facial information for authentication in a secure database. Experimental results confirm that the proposed system using EasyOCR is capable of recognizing number plates in Devanagari text, achieving an accuracy of 88.14% for English and Marathi number plates with a face recognition accuracy of 98.34% with dlib.