Conference Control System Based On Gesture Recognition

With the development of artificial intelligence technology,its related technology has gradually undergone a transformation from theoretical research to application,and is slowly moving from laboratory to the real market.In the daily operation of the company,it is often necessary for multiple departments to hold meetings and discuss projects.However,based on the traditional method,only a single speaker can make a presentation,which cannot allow other participants to comment on it,and it is difficult to carry out real-time discussion,which is not conducive to the efficient development of collaborative consultation.Based on the above background,and in view of the non-contact human-computer interaction problem of the current conference demonstration system,that is to say,real-time detection and recognition of gestures based on video stream can be realized through computer vision technology.This paper realizes a conference control system by analyzing some technologies.By using the method of deep learning,the user’s gestures in the continuous video input based on camera are detected and recognized,and the corresponding control signals are output,so as to realize the human-computer interaction of the conference presentation system.The system collects and identifies the continuous gestures of the controller through the camera.At present,it has completed five basic interactive functions such as click,pan,zoom,grab and rotate.The main work is as follows:The content of the pre-processing section of the image is the method of image denoising and gesture segmentation,the SNR can reach 25.8914 d B by using the improved method.In order to achieve better results,the image denoising method has been improved.To improve the accuracy of gesture recognition,a Softsign gesture recognition method combining the mixed attention mechanism and Softsign gesture recognition method is proposed,and a comparison experiment is conducted on both dynamic and static gestures,static gesture recognition rate is 97.02%,dynamic gesture recognition rate is 95.1%.On this basis,a meeting control system based on gesture recognition is designed.