“Hand Model” – A Static Sinhala Sign Language Translation Using Media-Pipe and SVM Compared with Hybrid Model of KNN, SVM and Random Forest Algorithms

The study of sign language recognition has been a thriving area of research for nearly twenty years. SSL translation using computer vision relies on extensive training using large number of images or video sequences. Yet, it poses a challenge for languages with limited resources, like Sinhala Sign Language. Due to the absence of accessible datasets for such investigations and lack of SSL recognition systems for real-time scenarios, the authors of this research have developed a completed system that consist of three components which are, static sign recognition, dynamic sign recognition and natural language processing to output a meaningful sentence. This paper describes the models and the results which are developed for static sign recognition. The final model for the Sinhala Static recognition has developed using media-pipe and SVM algorithm. And to compare the results and performance authors have used CNN, SVM, KNN and Random-forest algorithms to compare the results and performance. In this research, authors have made a valuable contribution by introducing a new dataset and developing a system for translating static SSL that gives 99.67% of accuracy and best performance when compared to the other two models and algorithms described in this paper.