Advances, Challenges and Opportunities in Continuous Sign Language Recognition

Abstract

Sign Language (SL) is the hands spoken language assisting the deaf to understand each other. Understanding SL by vocal people not only paves the way to contribute deaf and dumb in the workforce but also provides a fertile environment for analyzing the human motion and gesturing. Consequently, translating SL sentence into written or spoken language, known as Continuous Sign Language Recognition (CSLR) will help in integrating the deaf and dumb in the society. Most of the surveys in the field of Sign Language Recognition (SLR) spotlight on isolated SLR that mainly deals with words, numbers and letters each in separate. Moreover, these systems are designed to operate in artificial settings for the background, signer dependency and limited vocabulary. Even though for real-life CSLR is the objective, till now there is not a complete survey on CSLR that provides researchers with a comprehensive study on the advances, challenges and opportunities in this field. The presented piece of work analyzes the articles published earlier and illustrates the core stumbling blocks related to CSLR including: the dynamic hand detection and tracking, facial expression recognition, movement epenthesis detection and recognition methods as well as a comparative study on the available benchmark databases. An inventory of the applications which stand to benefit from CSLR are also brightened up. The conclusions and recommendations of this research can be a milestone for developing evolved and efficient CSLR systems.