Sign Language Recognition using Machine Learning

One of the ways to communicate with deaf and dumb individuals is through sign language. So, in order to speak with deaf and dumb people, one should learn sign language; yet, because not everyone can learn it, communication becomes nearly impossible. The goal of this study is to use machine learning to break through these communication hurdles. The majority of existing technologies rely on external sensors, which are out of reach for most people. We utilize OpenCV to take images and the CNN technique to train the machine, with the output being text. Many previous studies have offered methods for partial sign language identification, however this study intended for the full acceptance of American Sign Language comprises of 26 letters and 10 numbers. The majority of ASL letters are static, however few are dynamic. As a result, the goal of this research was to extract features from finger and hand motions in order to distinguish between static and dynamic gestures.