Gesture recognition based on lightweight neural network in virtual reality
Gesture is an important input mode in virtual reality environment,and the accuracy and efficiency of gesture recognition provide the basis for effective interaction in virtual reality environment.In order to improve the accuracy and speed of gesture recognition in virtual reality environment,a gesture recognition model based on lightweight optimized convolutional neural network is proposed.Firstly,feature extraction and initialization are carried out on gesture image samples,and a convolution neural network gesture recognition model is constructed.Then,with the help of MobileNet V2,the gesture recognition model parameter quantities and calculation amounts are effectively reduced through deep and point-by-point convolution.Finally,the parameters of the neural network are solved by minimizing the gesture recognition error.The experimental results show that the gesture recognition accuracy of the proposed lightweight neural network is slightly lower than that of the original convolutional neural network,but its recognition efficiency is greatly improved,especially in the scene of large-scale gesture sample recognition.The lightweight convolutional neural network shows high recognition comprehensive performance,and the gesture recognition adaptability is high in virtual reality environment.
万方数据
维普
