Preliminary evaluation of the Kinect V2 sensor for its use in virtual TV sets with natural interaction

A virtual TV set combines the real and virtual worlds to obtain an image in which the real elements give a sense of presence in a computer generated environment. One of the key elements for a credible mix is the interaction between the actors and the virtual world. In this paper, a preliminary study analyzing the viability of the Microsoft Kinect V2 sensor for its use in natural gesture detection in virtual TV sets, is presented. The learning of the system, through artificial intelligence techniques, of a series of natural gestures from the presenter, is proposed. This should allow the actor to create his own interaction with the environment reducing the training time needed before recording a TV show. Eight users with different levels of experience in virtual environments have created a series of intuitive natural gestures (using the Kinect V2) that have been used to evaluate the suitability of the sensor for this purpose, showing promising results.