High-Performance CNN-Based Facial Recognition on the Yale Face Database

Facial identification and detection have significantly risen due to deep learning techniques, especially Convolutional Neural Networks (CNNs), which perform better than traditional methods like Principal Component Analysis (PCA). CNNs automatically recognize strong features, giving better accuracy under numerous conditions, including illumination variations and stiff facial structures, unlike earlier systems that needed human feature extraction and contended with real-world differences. Using the Yale Face Database, 165 grayscale pictures of 15 persons were collected under varied lighting conditions and facial expressions; this research analyses Comprising three convolutional layers, max-pooling, and a fully linked dense layer, the suggested CNN model attained an astonishing 98% recognition accuracy. Rotation and horizontal flips among other data augmentation procedures expanded the variability of the dataset. With steady accuracy increases and continuous loss reduction over epochs, loss and accuracy curves showed effective learning. These findings illustrate the promise of this CNN model for uses including biometric authentication and security, where accuracy and adaptability are especially crucial. Future research proposes to integrate privacy-preserving technologies and solve demographic diversity, hence enhancing resilience against adversarial attacks and guaranteeing more general use.