Gesture recognition application with Parametric Hidden Markov Model for activity-based personalized service in APRiME

The paper introduces an approach to automatically recognize people's activity patterns within an "intelligent" building. We envisage a model of interactions with a smart phone and building. Various sensors in smart phone enable to recognize daily routine of people's activities automatically in building. The smart phone application; `Activity Pattern Recognition in Mobile Environment (APRiME)' recognized user's activity using location context as GPS and Wi-Fi. For increasing the accuracy of user's activity in the application, we suggest a new method to recognize hands movement using small parts of gesture. It is possible to recognize the actions user's activities via gesture. This paper summarizes some experiments that we performed using a smart phone that equipped with 3-dimension accelerometer to detect gestures. We can apply to Parametric Hidden Markov Model to learn and detect movement of gesture like as video analysis. This research will eventually be extended to realize an intelligent building like a `Big Brother', which knows everything you did using 3-dimensional accelerometer in smart building.