Research on Static Gesture Recognition Based on Deep Learning

Abstract With the continuous development and progress of times, ways human-computer interaction have become more diverse. In order to reduce spread new crown virus, gesture recognition has a hot topic in field recent years. Traditional is affected by environment database, etc., with poor robustness low accuracy. improve rate static gestures, this paper proposes establish deep learning model using CNN convolutional neural network, method based on template matching. By establishing palm diagram, image be recognized matched diagram feature point, rotated after matching, grayscale value again, so as extract part. Through experimental proof, algorithm can effectively rate, accuracy reached 93.17%, speed faster.