Research on Optimization of Static Gesture Recognition Based on Convolution Neural Network

Abstract

In order to solve the problems of low recognition rate and less recognition gesture categories caused by incomplete artificial feature extraction information in traditional static gesture recognition methods, a deep CNN framework is designed by using the principle of convolution neural network (Convolutional Neural Network, CNN) to recognize static gesture movements. Combined with a variety of optimal structures of convolution neural network in deep learning, a model with independent static gesture recognition function is realized. The model method can not only ensure the high accuracy and robustness of the recognition results, but also achieve the speed of smooth recognition.