Deep Learning based Indian Sign Language Recognition for People with Speech and Hearing Impairment

Languages that transmit meaning through the visual-manual modality as opposed to spoken words are known as sign languages. This visual language uses hand gestures and body postures when spoken communication is not possible or preferred. It is used by speech and hearing impaired individuals individuals to convey meaning. Therefore introducing a tech- nology capable of comprehending sign language could greatly enhance the social interactions and quality of life for deaf people. In contrast to ASL's single-handed gestures., ISL relies on bilateral hand motions to communicate. The suggested approach is convolution neural network(CNN) which translate the sign language into legible text. The collected data is preprocessed and used as input to CNN. We examined many cutting-edge architectures, including CNN, VGG16, and AlexNet, to tackle the issue of identifying Indian sign language. Among these archi- tectures, CNN architecture gave the best results. It is observed that the model gives 99.93% accuracy on test data. Tensorflow object detection is used in the suggested method to identify hand motions. With our system, the deaf community can communicate more easily and efficiently, without relying on external help.