• Purva C. Badhe
• Vaishali Kulkarni

The ability to communicate through gestural sign language is a fundamental human right for the hearing impaired community. It is the sole medium using which these people communicate with others. It is essential than to be able to converse effectively, the other person is aware of the gestures, their meanings and be able to understand them. However, in India, there has been negligence towards the development and awareness of this sign based language. The national standard sign language of India, known as Indian Sign Language (ISL) was recognized by mass only after 1970s and the standardized dictionary of ISL gestures is still in process of creation with more than 15000 words already released by Indian Sign Language Research and Training Center (ISLRTC) of Govt. of India. To improve the communication between ISL users and others who are not aware of ISL, recent technological developments can be utilized. There have been recent studies in the domain of developing such automatic systems that will translate ISL gestures into spoke languages. Some methods are sensor-based while others use computer vision. Various technologies employed include SVM, HMM, SIFT, DCT, KNN, ANN and so on. With emerging knowledge of artificial intelligence use of CNN and RNN is currently being employed in the same field. In this paper we are presenting one such methodology that makes use of hybrid CNN-RNN based approach to recognize dynamic ISL gestures into English language. A combination of CNN-RNN provides benefits of both spatial as well as hierarchical feature data. The proposed system is trained using self-created database and tested for 10 different classes with testing accuracy of 90% which is at par with other currently available methods.

Sign Language Recognition, CNN-RNN, Dynamic Gesture Recognition, Indian Sign Language, Transfer Learning.

"Dynamic Gestural Sign Recognition Using Deep Neural Network", International Journal of Emerging Technologies and Innovative Research (www.jetir.org), ISSN:2349-5162, Vol.10, Issue 8, page no.c762-c769, August-2023, Available :http://www.jetir.org/papers/JETIR2308290.pdf

"Dynamic Gestural Sign Recognition Using Deep Neural Network", International Journal of Emerging Technologies and Innovative Research (www.jetir.org | UGC and issn Approved), ISSN:2349-5162, Vol.10, Issue 8, page no. ppc762-c769, August-2023, Available at : http://www.jetir.org/papers/JETIR2308290.pdf