An Automatic Kiss Camera System Using Deep Neural Network Technique

Nowadays, there have been many entertainments to enrich our free time. Watching sporting events on the spot is one of the popular entertainments for people. However, the spectators do not interact much with the players most of the time. Moreover, spectators are sometimes occupied by their cellphones during the game. Although spectators get together in a stadium, they are actually farther apart from one another. To improve spectators’ interaction. Television directors use cameras to capture fun scenes and project them onto the screen. This scheme cannot only entertain the spectators but also increase the interaction between spectators and players. So far, this process is fully controlled by a cameraman. Therefore, it is required that the cameraman pays a lot of full concentration during games, especially at half-time. This job increases the cameraman's burdens. Hence, the purpose of this study is to automate this process using modern technology to reduce the pressure on cameramen and manpower. Besides, the advantages are high efficiency and flexibility to accommodate interactive experiences. The proposed scheme includes three stages. Firstly, facial recognition detects pairs of females and males. Secondly, a cellphone app collects all the detected couples, according to the distance between a female and a male. Users can choose one of those images with an interaction model. Finally, a spotlight will be aligned to the selected couple so that the cameraman can know where the camera should shoot for. A small-scale experimental system was implemented to demonstrate the feasibility of the proposed intelligent framework.