Comprehensive study of Sign Language Conversion Using Machine Learning

In this world, communication plays a major role to interact with each other. It acts like a bridge to convey message to other people. Human beings are able to manage their tasks by making use of their five senses, in which hearing is strongest among them. Some people have hearing loss from birth which results in physical health issues, which makes it difficult for them to acquire and say new words. These kinds of physically challenged people are known as Deaf and Dumb. The people who have hearing or speech impairments use the sign language to convey their messages. A Sign language is a combination of the hand gestures, face emotions and body movements. By using the sign language deaf and dumb people can communicate and deliver their message to others. There are several different varieties of sign language, primarily American Sign Language and British Sign Language. Normal individuals, however, struggle to understand sign language and the significance of others' actions since they are unable to use it. Our motivation for this study is to know how normal people can easily communicate with people deaf and dumb who understands only sign language. Here we analyses of various papers and their methods for solving this problem are discussed. The main solution of this situation is to create models which are helpful to understand the meaning behind the actions. So many models are created with various approaches like SURF with SVM and CNN, Neural machine translation, Deep LSTM, Natural language processing and other techniques. This study is done across various approaches to know the performance of each model their advantages and disadvantages in detail.