Finger spelled Sign Language Translator for Deaf and Speech Impaired People in Srilanka using Convolutional Neural Network

Abstract

Abstract: Sign language is a visual language used by people with speech and hearing disabilities for communication in their daily conversation activities. It is completely an optical communication language through its native grammar. In this paper, hoping to present an optimal approach, whose major objective is to accomplish the transliteration of 24 static sign language alphabet words and numbers of Srilankan Sign Language into humanoid or machine decipherable English manuscript in the real-time environment. Since Srilanka has a native sign language deaf/Signers become uncomfortable when expressing their ideas to a normal person which is why this system is proposed. Artificial Neural Networks (ANN) and Support Vector machines (SVM) have been used as the technologies of this proposed system. Pre-processing operations of the signed input gestures are done in the first phase. In the next phase, the various region properties of the pre-processed gesture images are computed. In the final phase, based on the properties calculated of the earlier phase, the transliteration of signed gesture into text and voice is carried out. The proposed model is developed using Python and Python libraries like OpenCV, Keras, and Pickle.