Hand Sign Recognition using Deep Convolutional Neural Network

Abstract

Sign language is the most complex language to understand by the end-user without knowing the meaning of the sign and it depends on the special gesture motion. The gesture marks are characterized by hands with aid by facial appearance and body position. In this article, Gesture recognition is proposed for static sign language using Deep Learning with image processing. The involvement contains dual solutions to the problem. One is resized with Bicubic static American Sign Language binary images. Besides that, good recognition results in of detection the borderline hand using the various edge detection techniques. Another solution is to classify the alpha characters of sign language using a Convolution Neural Network (CNN). Owing to the anticipated method, test accuracy of 96% and F1-Score of 99% have achieved for 10 different hand gesture classifications.