Comparative analysis of Bangladeshi sign language classification using transfer learning and fast fourier transformation

Abstract

Sign Language, a non-vocalized means of communication via the use of hand signs, gestures and motions to articulately express oneself and communicate with others for differently-abled people, especially hearing and speaking-impaired people. It is the most dominant form of communication in these communities. However, like all forms of communication, this too requires both sides to understand the language, which is the main barrier among many people of the community, due to not only having varied forms of sign language but also very few people outside the community who understands them. Our research aims to design a system using a glove or camera that reads Bangla Sign Language, processes and interprets it efficiently, and provides output in the form of Bangla Language for the other party to understand. We aim to focus on the Pre-processing aspect. Hence we decided to deploy Fast Fourier Transformation to decrease the pre-processing time.