A single depth silhouette-based hand gesture recognition for appliance interfaces in smart home environment

Abstract

The user-friendly and intuitive interface with household appliances is considered as one of highly promising fields for the researches in smart home environments. Instead of using traditional interface methodologies such as keyboards, mice, touchscreens, or remote controls for each appliance in smart home users can send commands to these devices by their hand gestures. This paper presents a system of hand gesture interface based on a single depth silhouette for appliance control in smart home. The core goal of our work is to create a user-friendly interface system which can identify specific hand gestures and use them to control appliances in smart home environments. To control appliances with hand gestures, we use both static and dynamic hand gesture model to generate control commands. In our methodologies, we first recognize the hand parts in a depth silhouette via the trained random forests (RFs). Then based on the information of the recognized and tracked hand parts, we recognize static or dynamic hand gestures. Appliance control commands are finally generated according to our implemented appliance interfaces. With this interface system, users can control some typical appliances in smart home such as TV, fan, lighting, security camera, doors and change channels, temperature, and volume by just hand gestures.