Hand Gesture Recognition Using Convolutional Neural Network

The human-machine interfaces will be more efficient when operated with voices or gestures without any hardware, such as mouse or keyboards. In particular, vision-based systems will be more appropriate in loud environments or environments with noises. In addition, the convolutional neural network has been applied more and more frequently in recognition problems with high accuracy such as face recognition, handwritten digits recognition, natural language processing. In this paper, we employed a convolutional neural network with 14 layers for the hand gesture recognition system with 6 different gestures of the right hand, and the input images were taken by a phone camera. The training data set was collected from the hand gesture of 7 people. The simulation results obtained using Matlab show that the system has an accuracy of 98.6% for hand images taken from the front with the appropriate brightness and suitable finger distance