A Transposed Hand Silhouette Interaction System Using Deep Neural Network

With the advance in new technology, the use of Virtual Reality (VR) or Augmented Reality (AR) techniques are successfully applied in the consumer market. Exhibition by hand silhouette is always an attractive application in many fields. A realistic exhibition in hand silhouette display is the most concern. However, it largely depends on the person's skill on hand exhibition and thus decreases the quality on the silhouette-based result. In this paper, we develop a transposed hand silhouette interaction system. To make a precise exhibition, we substitute the individual hand silhouette with a pre-defined hand silhouette template. It induces the important kernel for hand segmentation and classification which is used by the deep neural network. We build the whole system to make a precise and accurate hand silhouette exhibition in a transposed way. As a result, this innovative form of exhibition combined with technology is providing users with a fully interactive experience.