Journal of Research in Computer Science and Engineering ISSN: 2457-0818 (Online)

Authors: Al Mahmud Al Mamun, Md. Rakibul Hasan, Md. Mokhlesur Rahman, Md. Mahbubur Rahman, Sondip Poul Singha, Md. Yasin Khan

Abstract: Artificial Neural Networks is a computational model inspired by the structures and functions of biological neural networks. Artificial neural networks are used in all sectors of the recent world with its life-changing applications. In this paper, we review artificial neural networks. We specially focus on the history, architecture, learning technique of the artificial neural networks. Sign language is a visual communication system between hearing-impaired or hard of hearing people. The communication foundations are based on finger shapes, hand shapes, hand movements in space concerning the body, hand orientations, and facial expressions. In this paper, we have proposed a Deep Convolutional Neural Network to recognize the hand gestures. Our system uses the pre-processed images for training and is capable of extracting signs from video sequences under minimally cluttered and dynamic backgrounds using skin color segmentation. In our method, 44 signs of the five fingers are defined. The static images have been pre-processed as grayscale images and the model is trained with 100 numbers of images for each sign. Our proposed model has achieved an accuracy rate of 95.6 % for the five test images.

Keywords: Artificial Neural Networks, Deep Convolutional Neural Networks, Hand gestures, Sign language recognition