Multilevel Artificial Intelligence-Based Authentication System with Anti-spoofing Technique

Security concerns have long been a focal point in the realm of authentication systems, particularly within the context of single-level authentication systems. Password-based authentication, in particular, is recognized as vulnerable due to its susceptibility to bypass by malicious attackers. To strengthen security and enhance authentication processes, a multilevel authentication system incorporating biometric authentication is proposed. Among the biometric methods, face recognition stands out as a robust choice. To fortify the security of face recognition, the integration of face anti-spoofing measures is crucial to detect potential attacks such as printed image and replay attacks. To implement this multifaceted authentication system, the study employs convolutional neural networks (CNNs) as a machine learning method for both face recognition and face anti-spoofing. This approach enhances the system’s security and ensures the integrity of the authentication process. In summary, this study introduces a novel multilevel authentication system that combines traditional single-level password authentication with a secondary layer of face recognition authentication, fortified by face spoofing detection. The development, dataset generation, and model training are performed using the Visual Studio Code integrated development environment (IDE). Additionally, the system is structured within the web framework Django, providing a robust foundation for its deployment and usability. The proposed multilevel authentication system, combining password-based security with face recognition and anti-spoofing detection, demonstrates outstanding accuracy in training results despite a relatively small dataset size.