Face recognition plays a vital role in computer vision and biometric systems, particularly in home security applications where reliable human-system interaction is essential. It is widely used for security, identity verification, and human-computer interfaces. However, traditional face recognition techniques often struggle with factors like lighting variations, facial expression changes, and occlusions, which can significantly reduce their accuracy. To overcome these challenges, hybrid methods that integrate multiple recognition techniques have emerged, offering more robust and accurate performance across a variety of complex and dynamic conditions. This paper introduces a hybrid face recognition approach that integrates Convolutional Neural Networks (CNN), Fuzzy Logic, and Support Vector Machines (SVM) to achieve high accuracy and robustness. The proposed system combines the powerful feature extraction of CNNs, the precise classification ability of SVMs, and the uncertainty-handling capabilities of Fuzzy Logic. Specifically, the ResNet50 architecture is utilized to extract distinctive facial features. These features are then classified using SVMs, while Fuzzy Logic is applied to refine decisions, effectively managing ambiguous or imprecise inputs to improve overall system reliability. The approach is designed to enhance the performance of face recognition in dynamic home security environments by accurately identifying authorized individuals and granting them access. It demonstrates improved recognition accuracy under varying lighting conditions and facial orientations, ensuring consistent performance in real-world settings. Upon successful recognition, the system automatically unlocks the door, offering both secure and convenient access. By integrating advanced recognition techniques, the model reduces the likelihood of false acceptances and rejections, thereby bolstering security without compromising user experience. Experimental results indicate that the proposed hybrid model outperforms conventional methods in terms of accuracy, efficiency, and resilience. The evaluation further confirms that this integrated approach significantly enhances the robustness and adaptability of face recognition systems under real-world conditions. These outcomes underscore the potential of the proposed framework to serve as a foundation for developing more dependable and efficient face recognition solutions for practical deployment.
Chaabane et al. (Fri,) studied this question.