Abstract:
"
A Language is a way of communicating thought between two parties. Effective communication
exists when both parties actively engage in the session and respond. Just like speech and text that
we use to communicate, sign language is also a method used by the deaf and mute community all
over the world to communicate with each other and with other normal people. Not all of us are
aware of sign language and we do require a translation. As Sri Lanka is still known as a
developing country, most people don’t have a clear understanding of the sign language system
which is used locally. Therefore, many ordinary people are refusing to communicate with
disabled people. Only the people who engage with these communities are willing to learn sign
language. There are some basic translator systems that are used to translate sign language into
normal vocal language. But there are no robust interpreters available in Sri Lanka and most of
them lack real-time translation. The cost of the existing systems are high, and the quantity is not
enough for a wide communication system.
Through this report, we are proposing a solution to develop a translation system with a low resolution camera, which should be able to convert Sri Lankan sign language into English and
vice-versa in real-time. The research is divided into four major components. The first component
does the data acquisition and feature extraction part. Then it moves to the translation part where
extracted images are changed into letters. The third component will do the text and voice
assistant part and the final component will convert the texts into sign images. Our main goal is to
achieve a system that can perform in real-time. With the advancement of image processing and
machine learning, we think we can attain our goal.
"
