A Hybrid Face Recognition Scheme in a Heterogenous and Cluttered Environment

The recognition of human faces is an old application which can detect, track, and identify or verify human faces from a still or motioned picture stored digitally and is taken by digital camera. Though a significant development has already been seen in this domain, still some challenging issues are present which are yet to be addressed. Some of such issues are recognizing human faces in low intensity lighting condition, the presence of noise in images, scales, masquerade, etc. The present work focuses on using a conglomerate approach using deep learning algorithms with a support vector machine (SVM) to recognize faces. In the present paper, it has proposed a model to do facial image detection by using of deep neural network (DNN) with triplet loss function to extract features and support vector machine (SVM) classification algorithms to identify the images. The aim of the present work is to recognize the people from a variety of sources like videos, pictures and even sketches in a heterogeneous environment. When a selected image is fed to the system the model can be able to recognize detected faces. After that these detected faces are extracted and supplied to the recognizer which works with these faces and recognizes them as the respective persons. This recognition is done using SVM classifier which is very efficient when it comes to image processing. DNN and SVM have significantly improved the performance in recognition of faces in a cluttered and heterogeneous environment.