A Review Paper on Sign Language Recognition Using Machine Learning Techniques

Abstract

Correspondence is a significant piece of our lives. Hearing disabled individuals who can't talk experience various issues while bantering with regular people. These individuals locally can't speak with others without any problem. There are numerous manners by which individuals with incapacities attempt to convey. Sign language is one of the most natural and sensible ways for the disabled. Since deaf people cannot speak like normal people, they often have to rely on some form of visual communication. It has been shown that they sometimes find it difficult to communicate with ordinary people with their hands, as very few of them are recognized by most people. Communication through signing is a significant method for correspondence with the hearing impaired community. Gestures come in numerous structures, for example, hand gestures, two hands, and face movements. Gesture-based communication can be partitioned into two classifications: static and dynamic. Static signs are a hand setup and a particular shape, addressed by a single image. The dynamic sign is a moving action, represented by a sequence of images. Accordingly, the requirement for an intelligent computer-based program is in desperate need for deaf and dumb people that will empower them to convey all the more successfully with any remaining individuals utilizing their normal hands. This paper presents a top-to-bottom audit of Sign Language recognition techniques, a report related to those methods, and identifies challenges. The various methods and algorithms that have been used in sign language recognition projects have been refined and compared in terms of their advantages and disadvantages.