Building smart cameras on mobile tablets for hand gesture recognition

Mobile tablets have become very popular due to their portability and the wide diversity of the applications available. The touch screens and built-in accelerometers facilitate different forms of input instead of keyboards and mice. Nowadays, high-resolution cameras become a standard feature in a mobile device. Nevertheless, this camera is seldom considered to serve as a form of user inputs to the application, although similar technology is realized in some home entertainment systems. This paper describes our experience on making a smart camera on an iPad that can recognize pre-defined hand gestures. We study the time performance of performing image processing on an iPad. We find that due to the limited computational power of the mobile device, recognition results may not be available fast enough for a real-time application. We explore applying cloud computing to solve the problem. To the best of our knowledge, this is the first study on recognizing hand gestures on an iPad. Our results facilitate the development of a brand new type of applications that require smart cameras.