Journal of Emerging Trends in Engineering and Applied Sciences: Research Areas

Abstract:
Recognition of hand gestures in videos has found so many areas of applications such as: human-computer interaction, communication and computer systems controls. Many techniques and features have been reported in the literature for gesture recognition. However, these methods still need to be improved upon. In this work, a deep learning technique for recognition of static hand gesture using deep convolutional neural network is proposed. The model consists of two phases: image preprocessing and classification. The images are first processed by the convolution, maxpooling and RELU functions in the convolution layer of the CNN. The results of the convolution is then passed to a fully connected network layer to recognize the static hand gestures. The proposed model is implemented and tested on the publicly available ASL alphabet dataset from Kaggle repository. The model was trained using 69,600 images which consist of 29 gesture types. The model is then tested with another 17,400 images which consist of 29 gesture types. The experimental results show that the proposed system is able to recognize the static gestures with accuracy of 98.50 %.

Keywords: Gesture, Convolution, static. Deep Learning

Download full paper