Motion Based Indian Sign Language Recognition using Deep Learning
Atharv Ganpatye, Sunil B. Mane · 2022 2nd International Conference on Intelligent Technologies (CONIT) · 2022
Despite a large section of people being hearing impaired, there is still less awareness about sign language. These people use sign language to communicate with others, which includes various gestures using hands, facial expressions and body pose. There are many variants of sign language which vary from country to country. This paper proposes a model for Motion-Based Indian Sign Language (ISL) recognition using deep learning. The gesture is captured using OpenCV, key points are detected using MediaPipe and sign prediction is done by a trained LSTM (Long Short-Term Memory) model. The dataset used for this was created by the research team members. An average accuracy of 92% was obtained. The proposed system can be used for real-time Indian Sign Language recognition and can be integrated with video-conferencing applications.