A Review on Deep Learning Based Lip-Reading

Abstract

In the world of development and advancement, deep learning has made its significant impact in certain tasks in such a way which seemed impossible a few years ago. Deep learning has been able to solve problems which are even complex for machine learning algorithms. The task of lip reading and converting the lip moments to text is been performed by various methods, one of the most successful methods for the following is Lip-net they provide end to end conversion form lip to text. The end to end conversion of lip moments to the words is possible because of availability of huge data and development of different deep learning methods such as Convolution Neural Network and Recurrent Neural Networks. The use of Deep Learning in lip reading is a recent concept and solves upcoming challenges in real-world such as Virtual Reality system, assisted driving systems, sign language recognition, movement recognition, improving hearing aid via Google lens. Various other approaches along with different datasets are explained in the paper.