AI-Powered Secure Banking with Face & Liveness Verification

Abstract

Facial recognition has become a widely adopted biometric authentication technique due to its uniqueness and versatility. It plays a crucial role in identifying individuals in large crowds, making it a preferred method in security applications, automated surveillance, and missing person identification. Over the past four decades, face recognition has gained significant attention in computer vision research, leading to advancements in various algorithms. This paper presents a comprehensive review of facial recognition technologies, analyzing their strengths, limitations, and applications. It explores the fundamental concepts of face recognition, common methodologies, challenges, and potential future advancements.

A robust facial authentication system must not only recognize individuals but also detect spoofing attempts, such as those involving printed images or digital displays. One effective anti-spoofing measure is liveness detection, which examines facial movements such as eye blinking and lip motion. However, traditional liveness detection techniques are often ineffective against video-based replay attacks. To address this, this study proposes an AI-driven facial authentication system that integrates Convolutional Neural Networks (CNN) for facial recognition and liveness detection mechanisms. The system consists of two main modules: a CNN-based classifier for face authentication and an eye-blink detection module to assess natural facial movements. The CNN model is trained using publicly available datasets to enhance its effectiveness. These components are seamlessly integrated and implemented on an Android platform to develop a facial recognition application. Experimental results indicate that the proposed system successfully detects various spoofing attacks, including those involving masks, printed images, and digital screens.