Recent works in Sign Language Recognition using deep learning approach - A Survey

Across the globe, different countries have their own sign language used by hearing and speech-impaired people. The American, Arabic, British, Brazilian, and Indian Sign Languages have gained popularity and piqued researchers and engineers to develop solutions that will help in the task of Sign Language Recognition and Machine translation. Just like spoken languages, sign languages also have dialects, and hence multiple gestures can exist for one word. The task of Sign Language Recognition is complex and has been carried out for the last three decades. In this paper, we survey, compare, and summarize the proposed techniques of researchers in sign language and provide a proposed plan for our imminent work. An extensive survey of CNN networks, Transformer models, Sequence models, Keypoint-based methods, and Hardware-assisted methods applied on various sign language open datasets is presented in the paper which can form a basis of the literature survey for the research.