Development of CNN-based Virtual Mouse

Abstract

The technique of building a process of interaction between human and computer is evolving since the invention of technology. The mouse is a superb invention in HCI (Human-Computer Interaction) technology. Though wireless mouse technology is invented still, that technology isn't completely device free. A Bluetooth mouse has the need of battery power and connecting dongle. The proposed mouse system is beyond this limitation. This paper proposes a virtual mouse system supported HCI using computer vision and hand gestures. Gestures captured with a built-in camera or webcam and processed by a Convolutional Neural Network Model for classification among the desired mouse operations. The users are going to be allowed to regulate a number of the pc cursor functions with their hand gestures. Primarily, a user can perform left clicks, right clicks, and double clicks, scrolling up or down using their hand in several gestures. This technique captures frames employing a webcam or built-in cam and processes the frames to make them track-able and then recognizes different gestures made by users and perform the mouse functions. Therefore the proposed mouse system eliminates device dependency so as to use a mouse.