Reseacrh on the use of hand gestures as a medium of interaction between humans and machines is still intensively carried out by researchers to provide alternative studies of interaction media. This paper discusses the results of experiments with hand gesture input used for operations: click, doubleclick, drag, group (select more than 1 menu object). Deep Belief Network (DBN) algorithm plays a role in the process of recognition of hand gestures in the medium of interaction between humans and computers. The hand gesture recognition process consists of three stages, namely the segmentation of the hand area, the extraction of the characteristics of the hand gestures and the recognition of the patterns of the hand gestures. The gand gesture recognition process is carried out in real time based on human hand input. The experimental results show that the DBN method works quite qell and is quite fast in recognizing hand gestures that function as human-computer interaction media in real time interaction mode. Based on testing also known a pretty good level of accuracy related to the basic function of interaction, namely: click (84.3%), doubleclick (81.4%), drag (87.6%)
Keyword : Hand Gesture, Iacc, Kobe University, Deep Belief Network,