Live Sign Language Recognition: Using Convolution Neural Networks

Sign language is an effective way to communicate for the communities with speech and hearing difficulties. There has been a significant improvement in sign languages in recent years. Though the method is convenient, there is no consistent or uniform way for interpretation. This issue may be addressed using a computer aided system that facilitates hearing impaired persons to interact with others in the community. In India, the Indian Sign Language (ISL) has evolved over the decades to consider such issues. Here we discuss a vision-based hand gesture recognition method, as hands play crucial role in interpretation. The vision-based systems have their inherent limitations over conventional hardware methods, though improvements are possible through pattern and vision recognition methods. In this work, we consider Convolution Neural Networks (CNN) to arrange different signs into legible English text. The step-by-step methodology and the results are provided.