KARPALLAVI (HAND GESTURES) TO TEXT AND SPEECH CONVERSION: MARATHI SIGN LANGUAGE RECOGNITION

Abstract

In present research work ‘Karapallvi’ (Marathi Sign Language) converted into Marathi text. Karpallavi (MSL) is a hand gestures Language used for communication with individuals and hearing Speech impairments, allowing them to express their thoughts and ideas effectively. In Nashik district each linguistic community has its self-variant of hand gesture language. Instead of its significances, ‘Karapallvi’(MSL) requires rigorous research due to lacks of technological and digital support as compared to other sign languages. The ‘Karapallvi’ (MSL) consist ‘Mulakshare’(मुळाक्षरे), which lays down the foundation of Marathi script. Effective application of Deep Learning techniques gives automatic identification of these hand gestures. The research methodology contains Dataset Preparation, Preprocessing, Feature extraction, Model train and built techniques. It is found that the present research work not only increases the accessibility but also helps to preserve ‘Karapallvi’ (MSL) in the era of digital disruption. The ANN model gives (approx.91%) highest accuracy as compared to KNN, CNN, SVM.