The Institute of Infocomm Research (www.i2r.a-star.edu.sg/index.html) has developed a technology to help detect falls among the elderly. The system operates through a network of overhead video cameras which are strategically installed to continuously monitor areas such as hospital wards.

The fall detection system makes use of cutting-edge computer vision algorithms to monitor the activities of the elderly, eliminating the need for physical monitoring. It captures a number of reliable and tested attributes to model the typical traits of a fall. The system then detects when a fall occurs and immediately alerts the relevant parties.

Statistics have shown that about one-third of the elderly in Singapore, aged 60 years and above, have had recurring falls, and that fall-related injuries have serious repercussions. The elderly could suffer from severe physical injuries such as fractures, as well as psychological trauma due to the loss of self-esteem and fear of falling. Besides physical and mental damage, falls are also costly. Statistics from the United States have shown that the healthcare cost of falls and rehabilitation can come up to US$70 billion dollars a year.

The proposed architecture for the fall detection system.

The I²R team behind the fall detection technology comprises Principal Investigator Dr Huang Weimin, Dr Yu Xinguo, Dr Eng How Lung, Dr Panachit Kittipanya-Ngam, Mr Kok Tian Shiang and Dr Chiew Tuan Kiang. Their research has led to several patent-pending technologies. They are:

• The conversion of of 2D positions to 3D positions Recorded video footage will only give 2D information of a monitored patient who is in a 3D space. This technology will map 2D coordinates into 3D space, paving the way for more reliable detection of falls.
• Detection of 2D location and orientation information of the head and body Based on the 2D information and the 2D-3D conversion technology, the team developed the capability to detect and analyse sub-actions that could lead to a fall.
• Efficient camera calibration for fall detection The team has developed a tool that will allow fast and easy camera calibration.

The fall detection technology was showcased at the Silver Industry Conference and Exhibition (SiCEX) 2008 earlier this year. According to Dr Huang, the team is carrying out further improvements to the technology, in collaboration with companies and a nursing home. There are also on-going discussions with other healthcare organiations such as local hospitals and nursing homes to bring the technology to market, said Dr Huang. For more information on the fall detection technology, email inddev@i2r.a-star.edu.sg.