Search

In this paper, we present work in progress on activity recognition and prediction in real homes using either binary sensor data or depth video data. We present our field trial and set-up for collecting and storing the data, our methods, and our current results. We compare the accuracy of predicting the next binary sensor event using probabilistic methods and Long Short-Term Memory (LSTM) networks, include the time information to improve prediction accuracy, as well as predict both the next sensor event and its time of occurrence using one LSTM model. We investigate transfer learning between apartments and show that it is possible to pre-train the model with data from other apartments and achieve good accuracy in a new apartment straight away. In addition, we present preliminary results from activity recognition using low resolution depth video data from seven apartments, and classify four activities – no movement, standing up, sitting down, and TV interaction – by using a relatively simple processing method where we apply an Infinite Impulse Response (IIR) filter to extract movements from the frames prior to feeding them to a convolutional LSTM network for the classification. Financed by the Norwegian Research Council under the SAMANSVAR programme (247620/O70).

We present an evaluation of two approaches to the problem of inferring the presence of multiple persons in networks of binary sensors. This problem is critical for many applications of Ambient Assisted Living that benefit from knowledge of single- and multi-person presence where data is collected using ambient sensors. Both approaches make use of a graph representing sensors and their spatial relations. One approach uses a simple statistical method to derive a minimum number of people present, the other precisely tracks people through the sensor network. Both approaches are evaluated in a low and higher resolution setting on data of two persons inhabiting a laboratory equipped with motion sensors and contact sensors. Although the latter approach performs well tracking multiple persons, its inability to distinguish inactivity and absence make the former approach more suitable for this task, independent of resolution.

One of the most appealing applications of wireless sensor networks (WSNs) is in human detection and tracking. The aim of these applications is to detect if a person is in an area of interest, and to keep track of his location at every instant of time. In recent years, we have seen a growing interest in the development of proposals for the use of WSNs in detection and tracking applications for healthcare. In the particular case of a patient suffering from dementia, it is very important to detect him and keep track of his location at every time, to avoid that the patient may enter to a zone of risk without supervision. When an event of interest is detected, such as wandering, an action may be taken by sending out a notification to the caregiver personnel. In this chapter, we review the most important proposals regarding the use of WSNs for human detection and tracking in healthcare applications. Moreover, we introduce a model for detection of patients suffering from dementia, based on a WSN that uses binary sensors. The proposed model is able to detect if a patient leaves a secure zone without supervision, and to emit alerts directed to caregivers.