Many elderly people are in need of care. However, the care in special facilities is expensive and a 24/7 support is not always possible, nor is it always needed. To allow the elderly to live in their own home longer, while giving them the needed prospect of security and the availability of help in moments when it is really needed, home assistance systems are being developed. In this project we focus on cases when a person falls down and cannot get up by himself/herself.
The goal of this project is a reliable and automated computer vision system to enable an independent lifestyle for the elderly and disabled. In contrast to other projects, the system relies solely on computer vision techniques. The daily lives of the people involved will not be affected and it overcomes many problems of existing systems. Main goal is the robust fall detection. Since the detection set-up consists basically of IP based cameras and a central storage and calculation server, the system is open and flexible towards other applications: Fire, smoke and water detection and assistance for medication can be introduced in a second step.

Motivation

Every year, an estimated number of 1000 persons aged 60 and above are dying in Austria because they collapsed in their homes. More than 100.000 people have to be treated in hospitals after fall accidents.  Research found that half of those patients with a “long lie” (i.e.  those remaining on the ground for more than one hour after a fall) died within six months of the fall, even if there was no direct physical injury (death was usually a complication  such as bronchopneumonia, dehydration or hypothermia). Even if the fall didn’t cause any physical injury, the impact on the psyche should not be underestimated. In many cases such accidents represent the first step to loosing independence for elderly people. 30% of people older than 60 years live alone. This project can take care of the communication with remote assistants in case of an accident, thereby prolonging the possibility of independent life of the elderly.
For more details and further statistics look at the preliminary research project COSAMED here.

Scope of this project

The project will develop a closed system for the automated event detection and the communication with mobile devices. It is based on the idea of “smart homes”, but uses state of the art 2D and 3D computer vision techniques. To make the system economical, low prize cameras are used.
Today’s state of the art event detection in assisted living is an alarm button on a wristband. Such an approach has many drawbacks: It needs to be worn all the time (and in reality this is rarely the case) and it needs human activation in case of an accident – which is not possible in all cases, e.g. loss of consciousness – or if a mechanical fall detection sensor is included in the band the false alarm rate is very high.
Cameras are passive, multifunctional sensors with a wide spectrum of possible usage. They are non-intrusive, relatively reliable and can be used 24/7. Using video signals, the system’s functions are open to updates or adding of new analysis methods (like e.g. fire/smoke detection) without changing the hardware. Furthermore, the retrospective analysis of summary statistics, tracking trajectories and where appropriate image data, could be used to provide valuable insight into behavior and health of a person.

In this project technical expertise of the state of the art computer vision is merged with the needs of well known consumer carriers in the field of assisted living.

Preliminary research project COSAMED

The project coordinator, the CogVis GmbH, already did a preliminary research project for medical care called COSAMED. The feasibility of such a fall detection system has been assessed during the COSAMED project, interviews with both consumers and demand carriers have been carried out, and the results were promising.

Details on the COSAMED project, statistics, background, related work and preliminary results can be found here.

 

This project is supported by:

Bundesministerium_für_Verkehr,_Innovation_und_Technologie_(BMVIT)

Österreichische Forrschungsförderungsgesellschafft mbH (FFG)