Hand State Combination as Gesture Recognition using Kinect v2 Sensor for Smart Home Control Systems

Smart home technology is evolving along with advanced technology evolvement. One of the technology which is constantly evolving is a human-machine interface. Within this group, a gesture-based interface enables to change this movement becomes a command for computer systems without contacting any surfaces. A gesture is one of the most convenient and natural ways to control a smart home. This research discusses the development of hand gesture recognition using hand state combinations to control various electronic devices for a smart home to facilitate human activities in a home. The system is built by performing hand gesture recognition from a command received by Kinect v2 sensor with hand tracking process to get the desired request based on existing hand state combination. There are three hand states with nine hand state combinations. Each combination means different commands for different electronic devices. This study uses the hand state from Kinect v2 dataset, which are open, close, lasso, untracked, and unknown. Based on the experimental results, the system accuracy reached 87% with an optimal distance of three meters.