Multimodal authentication system with facial recognition and totp for adaptable secure access

• Autores: Mecola Bernedo, Jesús Christopher, Tirado Ávila, Julio David, Mendoza de los Santos, Alberto Carlos
• Formato: artículo
• Fecha de Publicación: 2026
• Institución: Universidad La Salle
• Repositorio: Revistas - Universidad La Salle
• Lenguaje: español
• OAI Identifier: oai:ojs.revistas.ulasalle.edu.pe:article/339
• Enlace del recurso: https://revistas.ulasalle.edu.pe/innosoft/article/view/339; https://doi.org/10.48168/innosoft.s29.a339; https://n2t.net/ark:/42411/s29/a339
• Nivel de acceso: acceso abierto
• Materia: Biometric authentication; facial recognition; Multi-modal systems; two-factor authentication; usability; biometría; reconocimiento facial; sistemas multimodales; autenticacion de doble factor; usabilidad

The increasing sophistication of cyber threats has revealed the limitations of password-based authentication mechanisms. Although multifactor authentication (MFA) has emerged as a security standard, traditional MFA schemes often impose rigid verification flows that negatively impact usability and system adoption. This work presents the design, implementation, and evaluation of a flexible multimodal authentication system that enables user verification through facial recognition or a time-based one-time password (TOTP), in combination with a conventional password. The system was developed in Python following a Model–View–Controller (MVC) architecture to ensure modularity, maintainability, and scalability. The biometric module integrates OpenCV and the face_recognition library to extract and validate facial embeddings, while PyOTP enables TOTP generation and verification under the RFC 6238 standard. Experimental results demonstrate a biometric accuracy of 85%, an average authentication time of 2.1 seconds, and a False Acceptance Rate (FAR) of 0.8%. Meanwhile, TOTP validation achieved a 94% success rate. These results demonstrate that a flexible OR-based MFA approach can balance usability and security, making the system a viable alternative for academic environments, research prototyping, and low-infrastructure scenarios that require secure yet user-friendly identity verification mechanisms.