Deep Learning-Based Face Recognition Using Convolutional Neural Networks: A Comparative Analysis with Real-Time Application

This paper presents a comparative study of deep learning-based face recognition using Convolutional Neural Networks (CNNs) and traditional machine learning approaches. The proposed system is designed to improve facial recognition accuracy while supporting real-time applications. The study evaluates a CNN architecture against Eigenfaces (PCA), Fisherfaces (LDA), and Local Binary Patterns (LBP) using the Labeled Faces in the Wild (LFW) dataset. The proposed model combines convolutional layers, ReLU activation, max-pooling, fully connected layers, and dropout regularization. The experimental results reported in the paper demonstrate a recognition accuracy of 98.25% for the proposed CNN, compared with 85.12% for the Eigenfaces baseline. The study also presents an OpenCV-based prototype for real-time face recognition and evaluates computational performance across different hardware configurations. Additionally, the paper discusses architectural optimization, data augmentation, model robustness, and opportunities for future improvements in edge computing, adversarial security, and biometric authentication.