Naslov
Activity Detection in Smart Home Environment

Autori
Skočir, Pavle ; Krivić, Petar ; Tomeljak, Matea ; Kušek, Mario ; Ježić, Gordan

Izvornik
Procedia Computer Science (1877-0509) 96 (2016); 672-681

Vrsta, podvrsta i kategorija rada
Radovi u časopisima, članak, znanstveni

Ključne riječi
event recognition ; activity detection ; data filtering ; Internet of Things

Sažetak
Detection of human activities is a set of techniques that can be used in wide range of applications, including smart homes and healthcare. In this paper we focus on activity detection in a smart home environment, more specifically on detecting entrances to a room and exits from a room in a home or office space. This information can be used in applications that control HVAC (heating, ventilation, and air conditioning) and lighting systems, or in Ambient Assisted Living (AAL) applications which monitor the people's wellbeing. In our approach we use data from two simple sensors, passive infrared sensor (PIR) which monitors presence and hall effect sensor which monitors whether the door is opened or closed. This installation is non-intrusive and quite simple because the sensor node to which sensors are connected is battery powered, and no additional work to ensure power supply needs to be performed. Two approaches for activity detection are proposed, first based on a sliding window, and the other based on artificial neural network (ANN). The algorithms are tested on a dataset collected in our laboratory environment.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo

POVEZANOST RADA