Hybrid CNN and Vision Transformer-Based Multi-Factor Authentication for Enhanced Security in Online Examination Systems

With the rapid growth of online examination platforms, maintaining high levels of security, integrity, and user authentication is paramount. While existing methods utilize traditional security measures, the integration of advanced deep learning techniques can further improve the robustness of these systems. In this paper, we proposed a hybrid model that combines Convolutional Neural Networks (CNN) and Vision Transformers (ViT) for facial recognition in a multi-factor authentication system for online examinations. The CNN model captures local feature patterns, while the ViT captures the global relationships within the image, by combining these architectures can provide highly effective method for facial recognition tasks. The proposed model use both CNN and ViT in feature extraction and provides robustness. From the experiment results is observed that the proposed model achieves an accuracy of 98%, outperforming conventional methods in the domain of online examination security.