One of the best challenges in Lip Reading is to apply this technology in embedded devices. In current solutions the high use of resources, especially in reference to visual processing, makes the implementation and integration into a small device very difficult. In this article a new and efficient algorithm for detection and tracking of lips is presented. Lip Finding and Tracking are customary f...

This paper addresses the problem of lip segmentation in color space, which is a crucial issue to the success of a lip-reading system. We present a new segmentation approach to lip contour extraction by taking account of the color difference between lip and skin in color space. Firstly, we obtain a lip segment sample via a color transformation sequence in 1976 CIELAB and LUX color spaces. Second...

This paper studies various deep learning models for word-level lip-reading technology, one of the tasks in supervised video classification. Several public datasets have been published research field. However, few investigated techniques using multiple datasets. evaluates four publicly available datasets, namely Lip Reading Wild (LRW), OuluVS, CUAVE, and Speech Scene by Smart Device (SSSD), whic...

Recently, lip reading has become one of the most important fields study in field artificial intelligence. In this study, process was performed Turkish language using convolutional neural networks (CNNs). For purpose, people were asked to record numbers video (61 video), and 9 also collected from YouTube. The dataset for 20 numbers. only used sounds completely removed. Due small dataset, it trie...