A Secure Bank Locker Access System using Facial Recognition with Liveness- Based Spoofing Detection

Traditional security mechanisms for bank lockers, such as physical keys and PIN codes, are increasingly vulnerable to theft and unauthorized access. This paper presents the design and implementation of a robust, modern security system that leverages biometric technology to address these vulnerabilities. The proposed system utilizes facial recognition for user authentication, enhanced with a critical liveness detection module to prevent sophisticated spoofing attacks. The system architecture is built on a foundation of computer vision and deep learning techniques, implemented using Python and the OpenCV library. The core functionality involves capturing a user’s live facial image, verifying their identity against a pre-enrolled database, and simultaneously performing a liveness check to ensure the subject is a real person and not a static photo or video. The liveness detection model, a Convolutional Neural Network (CNN) named LivenessNet, is specifically trained to distinguish between real and spoofed facial images by analyzing subtle micro-textures. The entire system is integrated into a user-friendly application with a clear workflow for user login, authentication, and access control. This paper details the system’s functional and non-functional requirements, architectural design using UML diagrams, and the implementation of the core algorithms. The result is a highly secure, convenient, and reliable alternative to conventional locker security systems, setting a new benchmark for protecting valuable assets in the banking sector.