Classification of Hand Gesture in Indonesian Sign Language System using Naive Bayes

This paper proposes the use of Naive Bayes to classify hand gesture in Indonesian Sign Language System (SIBI). The proposed system can be used as a training method for normal people to learn sign lenguage so they can overcome obstacles in communicating to people with hearing disability. Hand gesture is captured using a low cost and portable finger motion capture device called Leap Motion. Hand gestures of 10 sign words are examined and data is extracted to obtain about 19 features. Our experiments deliver good results with accuracy of 80.5% to the trained data from ideal environment and 70.7% to the instant untrained data. Hence, Naive Bayes can be used to classify hand gesture in Indonesian Sign Language System captured using Leap Motion.