Hybrid techniques based on deep learning and mathematical approach for lips movements detection

Abstract

Lips movement make new approach for human relation ships specially for hearing impaired persons, so a lot of researches are done to make service for them via a lot of IOT application. In this research deep learning was adopted for tracking lips movements for speech understanding. In addition to the deep learning a mathematical approach are applied via curve fitting using polynomial equation with Euclid's distance which applied in different words and different genders. The proposed algorithm show the hybrid approach has more benefit compared with traditional technique. The data set used its locally generated with special Arabic selected word.