An Efficient Real-Time Recognition of Static Kannada Sign Language

Sign language is a gestural language used by speech and hearing-impaired people to communicate with each other and with other hearing individuals. Sign language is often used in educational settings to facilitate learning for deaf students. Sign languages are distinct from spoken languages and differ from one country or region to another. The system proposed seeks to fill the void by applying machine-learning techniques for the real-time recognition of Static Kannada Sign Language which is Vyanjangalu. The self-constructed comprehensive dataset, comprises 3,400 images representing 34 static signs of Kannada Vyanjanagalu. Utilizing Mediapipe as a feature extractor, this research explored the efficacy of nine distinct models for Kannada Vyanjanagalu recognition. The standout performer proved to be the Gated Recurrent Units and Support Vector Machine achieving an outstanding accuracy rate above 99%. The technology is adept at recognizing complex signs, handling multiple signers, and facilitating translation to the Kannada language. The integration of cutting-edge technology into this project addresses most of the research gaps and promises to significantly improve the quality of life and communication for the hearing-impaired population, emphasizing the importance of addressing their unique needs.