Transfer Learning Based Facial Recognition System

Facial recognition has become an important factor in our life because of its usage in various applications. These applications include unlocking smartphones, smart advertising, prevent retail crimes; tracking school attendance, recognizing important dignitaries at events, find missing persons and others. Apart from its uses, it also has various challenges to encounter such as different poses, background, noise, illumination and others. We have various Convolutional Neural Network (CNN) models and learning methods employed which gives optimal results. These models have various layers in them which extract important information from the input image, reduce dimensionality and pass this information to further layers. This paper focuses on two aspects facial detection and classification using deep learning. The work proposes a deep learning model to capture these aspects. Experimentation takes images from the database, applies Image Generator for Augmentation, trains the model for a certain number of epochs where the model adjusts its weights using backpropagation to give better accuracy. Performance evaluation of the model shows that it has been successful in predicting the faces of individuals correctly with an accuracy of 89%.