Face recognition in constrained unconstrained video Sequence using deep network

Abstract

As the amount of data being generated online increases as well as adherence to identity newlineverification systems the facial recognition technology has recorded significant developments in the past few years. It is very useful in activities like policing, investigating, and security because it offer sound method of confirming or recognizing an individual by the look of his or her face. newlineAs the research work is focused on face recognition from videos using deep learning, a new modified CNN model is proposed for this task. The first part of this research discusses different deep learning models including CNNs and other combinations that has performed very well in face recognition and matching. These models are described in detail to expose the specialties and shortcomings of each one. These insights lay the foundation for the second phase of this work, which proposes a new approach to Facial newlineRecognition based on the training on big data set. The effectiveness of the proposed approach is assessed with two benchmarks: the Labeled Faces in the Wild (LFW) and the YouTube Faces (YTF), showing superior recognition rates. This research aids the development of better accurate and reliable facial recognition technologies and gives direction for future research.