Intelligent System for Multi-Layer Lip Reading Based Microlearning

Abstract

     Intelligent systems can be used to build systems that simulate human behavior. One such system is lip reading. Hence, lip reading is considered one of the hardest problems in image analysis, and thus machine learning is used to solve this problem, which achieves remarkable results, especially when using a deep neural network, in which it dives deeply into the texture of any input. Microlearning is the new trend in E-learning. It is based on small pieces of information to make the learning process easier and more productive. In this paper, a proposed system for multi-layer lip reading is presented. The proposed system is based on micro content (letters) to achieve the lip reading process using deep learning and auto-correction models for error detection and correction of the accumulated words spelled by the lip reading system.The proposed system was implemented on a proposed dataset gathered using 20 persons of different ages and genders. The model achieved 98% accuracy in micro content recognition and 73.6% accuracy in accumulated word lip reading.