Camera-based Korean Sign Language Recognition for Banking Transaction

Gesture recognition has improved performance with the development of deep learning technique. In this paper, we were set the basic scenario to go to the bank and do the banking between the bank teller and customer who is a deaf person with automatically translate the banking by gesture recognition. For gesture recognition, we have defined gestures such as deposit or withdrawal and balance handling related to banking, and built a related database. We recognized gestures using CNN, and as a result of experiments, we found future research topics such as split the continuous motion, adaptation to changes in skin color, and precise recognition between similar motions