Sign Language Recognition Using Deep Learning-A Review

Abstract: For millions of people who are deaf or hard of hearing, sign language is more than just a communication tool it is the very foundation of their identity and daily life. Yet, most of the hearing world cannot understand it, which creates a serious and ongoing barrier to education, employment, and basic social participation. This paper looks at how deep learning can help close that gap through automated Sign Language Recognition (SLR). We reviewed systems that range from simple rule-based approaches all the way to the latest transformer models and graph neural networks. To make sense of all these different methods, we grouped them into a four-tier classification based on how advanced and how deployable they are. We then studied ten important research papers published between 2015 and 2024, compared them across factors like accuracy, speed, and real-world usability, and identified six gaps that still need to be addressed — including the near- total absence of Indian Sign Language datasets and the difficulty of running these systems on everyday mobile devices. Based on all of this, we outline a practical recognition framework called the Deep Sign Recognition Framework (DSRF) that aims to work in real-world settings, support Indian Sign Language, and run on standard hardware without needing expensive equipment.

Keywords: Sign Language Recognition, Deep Learning, CNN, LSTM, Transformer, Hand Gesture, Computer Vision, Accessibility, Indian Sign Language, Transfer Learning.