Human Activity Recognition by Using Privacy-conscious Sensor System

DOI Web Site

Bibliographic Information

Other Title
  • プライバシーに配慮したセンシングシステムによる人間の行動認識

Description

<p>In these days, Japan is facing a declining birth rate and an aging population. In this situation, watching over the elderly single adults has become a big issue. To deal with this problem, we have developed the human activity recognition (HAR) system which is using privacy-conscious sensors. Here, the “privacy-conscious” means avoiding the use of the cameras and the microphones. This system is composed of several kinds of sensors; the human sensor, the temperature and humidity sensor, the illuminance sensor and the door open/close sensor. In addition, we developed human activity logging system to record daily activities. To recognize human activity, we analyzed the data collected from these systems by utilizing the machine learning. As a result, we found that the HAR by using the supervised machine learning can achieve 0.9 accuracy. In this case, the human activities are the meal, relax, work, out go, face wash, cleaning, cooking, washing, sleep and others. In addition, we also found that the HAR by using unsupervised machine learning can achieve 0.9 accuracy. In this case, the human activities are the relax, out go, sleep and others.</p>

Journal

Keywords

Details 詳細情報について

Report a problem

Back to top