ENHANCING THE PERFORMANCE OF SIGN LANGUAGE RECOGNITION USING CNN AND RNN CLASSIFIER

ENHANCING THE PERFORMANCE OF SIGN LANGUAGE RECOGNITION USING CNN AND RNN CLASSIFIER

Sai Sampath Gunupuru1 , Akshidha2 , Pragna3 , Uma Sankara rao V 4, Aditya 5

1,2,3,4,5Department of Computer Science and Engineering, GST, GITAM University, Visakhapatnam, AP, India.

Abstract:

Sign language gave hope to deaf and dumb people to communicate with others. Although there is a significant development in sign language, Sign recognition systems are created to increase efficiency in sign recognition. Our sign language recognition technology was created to aid in communication for individuals suffering from hearing and visibility issues. We have created it using CNN (Convolutional neural network) and RNN (Recurrent neural network). After pre-processing the images, the images are used in the dataset containing the 25 English alphabets. We built CNN and trained it with the dataset. After training the model we have predicted the accuracy. To increase the accuracy we have enhanced the CNN model with RNN. The best accuracy obtained is 88.8% for CNN with RNN.

Keywords- Convolutional neural network, Recurrent neural network