A Static Hand Gesture Based Sign Language Recognition System using Convolutional Neural Networks
Diksha Hatibaruah, Anjan Kumar Talukdar, Kandarpa Kumar Sarma
Abstract
Sign language (SL) is the basic means by which deaf and dumb person communicate. However, these SLs are not known to the majority of healthy people. In our paper, we have developed a SL interpreter that takes the input sign gesture and gives the output in a display device. We have used Convolutional Neural Networks (CNNs) to train the system with a given database. For our work, we have used Indian SL database, consisting of 26 alphabets along with 10 digits. We have applied Histogram BackProjection technique for segmentation of images. These datasets are then formed into classes which are fed to a CNN for training and testing. After training we find out the testing accuracy of 99.89% and validation accuracy of 99.85% at 5 epoch. After this, we test the system with real time input and find out the result in the display device.