This paper is based on gesture recognition. It aims to transform gesture signal into electronic communication with home facilities. The main procedure consists of following steps: First, Curie chip collects the command gesture. Then, the gesture will be translating into accelerometer and gyroscope information. The Curie chip will integrate them into a motion vector and transport them to the k-means neural network as training. After training, each gesture represents a command attached to a facility control, and gesture control is revealed for smart home. Additionally, the whole system can be integrated as a wearable device, which relives complex button controller and improve portability. In brief, the research aims to replace remote controller and create a more convenient and pleasant living environment for users.