Prototype of Indoor Activity Estimation System with Low Load
Shintaro Imai, Yusuke Kishikawa, Yoshikazu Arai, Toshimitsu Inomata · 2018
In this research, we aim to estimate observed person's indoor activities with low load. The proposed system satisfies the following three functional requirements i.e., (F1) an observed person is not required wearing tags/sensors, (F2) not using camera, and (F3) using inexpensive sensors. The system estimates the observed person's activities based on acquired data from human sensors, power consumption sensors, illuminance sensors and water flow sensors. From the result of preliminary experiment, the system uses neural network for estimation. We implemented a prototype system and confirmed that moderate estimation accuracy and low load are compatible.