Deep Learning Based Hand Gesture Recognition for Emergency Situation: A Study on Indian Sign Language
Qazi Mohammad Areeb, Mohammad Nadeem
Abstract
Sign Language is used to convey feelings and thoughts, as well as to reinforce information given in everyday discussions. The goal of Sign Language recognition is to recognize and comprehend important human body gestures. Deep learning is a subset of machine learning that has lately gained traction in the recognition of sign languages. The current research focuses on how deep learning may be used to solve the challenge of identifying hand gestures in a collection of videos for emergency situations. To feed the model, a number of frames were taken from the videos. A pre-trained VGG-16 and a recurrent neural network with a large short-term memory make up the model (RNN-LSTM). The model achieved an accuracy of 98% on an Indian Sign Language Dataset of Hand Gestures for Emergency Situations. Deaf people can use sign language as a kind of emergency communication to help them deal with these circumstances. In this study, sign recognition could be utilised to address circumstances like pain, calling for help, or having to see a doctor.