Search

In recent year, lip-reading systems have received much attention, since it plays an important role in human communication with computer especially for hearing impaired or elderly people. In this paper, we introduce a new visual feature representation combines the Hypercolumn Neural Network model (HCM) with Hidden Markov Model (HMM) to achieve a complete lip-reading system. To check our system performance we introduce the Arabic language to it. According to our knowledge, this is the first time that a visual speech recognition system is applied for Arabic language. Experiments include different Arabic sentences gathered from different native speakers (Male & Female).

In this paper, we present a face alignment approach using granular features, boosting, and an evolutionary search algorithm. Active Appearance Models (AAM) integrate a shape-texture-combined morphable face model into an efficient fitting strategy, then Boosting Appearance Models (BAM) consider the face alignment problem as a process of maximizing the response from a boosting classifier. Enlightened by AAM and BAM, we present a framework which implements improved boosting classifiers based on more discriminative features and exhaustive search strategies. In this paper, we utilize granular features to replace the conventional rectangular Haar-like features, to improve discriminability, computational efficiency, and a larger search space. At the same time, we adopt the evolutionary search process to solve the deficiency of searching in the large feature space. Finally, we test our approach on a series of challenging data sets, to show the accuracy and efficiency on versatile face images.

In this paper, we propose a supervised learning method in Hyper-Column Model (HCM). HCM is a model to recognize images, and consists of Hierarchical Self-Organizing Maps (HSOM) and Neocognitron (NC). HCM complements disadvantages of HSOM and NC, and inherits advantages from them. There is a problem, however, that HCM does not suit general image recognition in HCM since its learning method is an unsupervised one with competitive learning which is used by Self-Organizing Map (SOM). Therefore, we extended HCM to a supervised learnable model with an associative memory of SOM. We have found that an ability of HCM with supervised learning is superior to the one with unsupervised learning.