Abstract

In this paper, I would like to present experimental results facilitate hand gesture recognition and classification based on machine learning algorithm and computer vision techniques. This is accomplished by two subtasks. First, computer vision techniques are used to detect and track a hand, identifying key features such as the location of each fingertip and the center of its palm. The system implements a Support Vector Machine to both train and classify gestures. By using both computer vision techniques and artificial intelligence, enough knowledge is made available to the system to support human-machine interaction via hand-gestures. I developed a system that gathers the aforementioned knowledge using frames captured by the webcam and wrote a host of applications that use this knowledge such as video-playback control and gesture-based drawing.