RESNA 27th International Annual Confence

Technology & Disability: Research, Design, Practice & Policy

June 18 to June 22, 2004
Orlando, Florida

A Computer-Vision System To Detect Falls In The Home: Preliminary Design and Development 

Tracy Lee, M.A.Sc and Alex Mihailidis, Ph.D. P.Eng.
Intelligent Assistive Technology and Systems Lab, Department of Occupational Therapy, University of Toronto, Toronto, Canada


We are developing a home-based emergency response system that can automatically detect falls and other emergency situations in order to support safe and secure aging-in-place. Based on the principles of ubiquitous computing, the system will detect when an emergency situation has occurred and prompt the user to ask if they are okay. ‘Listening’ to the occupant’s response, the system will intelligently decide upon the proper course of action and carry it out. This paper will present the fall detection component of the proposed intelligent emergency response system (ERS) and some preliminary results.


Fall detection, dementia, computer vision, emergency response, aging-in-place


Falls are currently one of the leading causes of morbidity and mortality in adults over the age of 65, with increased risks associated in adults with dementia [(1), (2)].It is estimated that one in three older adults will experience a fall over a one-year span, with one-third of these falls occurring in the home (3). Furthermore, fall incidence in older adults with dementia increases to 70-80% of the dementia population, which is approximately twice the incidence of falls in cognitively intact older adults (2). Providing immediate response and care when a fall occurs is a key concern, and is becoming increasingly difficult to accomplish as more older adults choose to remain in their own homes, often alone. Situations have been reported of the person left lying on the floor after a fall for an extended period of time before receiving assistance. This ‘long-lie’ often drastically reduces the probability of recovery (and survival) of the person (4).

Worn fall detectors (e.g. Tunstall Group Ltd., and emergency response systems (ERS) (e.g. Lifeline Systems Inc., are some examples of currently available commercial technologies that attempt to address the problem of fall detection in the elderly population. Worn fall detectors are mechanical sensors that are worn on the hip and triggered when both the orientation and acceleration forces of the person reaches a pre-set threshold. A common form of ERS is a telephone-based personal system consisting of the person wearing a small help button as a necklace or wristband, which the person pushes manually when an accident has occurred, and a two-way telephone system that connects the user to emergency services. A primary limitation of these devices is that they require effort from the user in order to be effective. For example, the user must remember to wear the device, which cannot be reliably assumed in older adults with dementia. Furthermore, if a fall occurs causing serious injury, the user may not be capable of pushing the button, thus rendering the device ineffective.

To overcome some of the difficulties that the dementia population experiences with aging-in-place and with currently available devices, we are developing an intelligent ERS that can automatically and confidently detect if an emergency situation has occurred, such as the person becoming ill or falling, and then subsequently calls for appropriate assistance.


The goals of this research is to develop a simple, cost-efficient, and robust system which can reliably detect falls in the home and provide a safe and secure environment for aging-in-place. Certain design criteria for the system were established, such as:


The proposed intelligent ERS can be broken into three modules, namely the sensing module, the prompting module and the planning module. The sensing module is responsible for visually tracking the occupant in his/her environment and confidently detecting a fall. The prompting module is activated when a fall has been detected and will ask the person if everything is okay, ‘listen’ to the response and then determine the appropriate course of action to take. Finally, the planning module will increase the intelligence of the system to intrinsically recognize areas of acceptable inactivity (i.e. the bed, or sofa) and eventually learn the living patterns of the occupant. Other research groups are currently investigating different methods using computer vision and other sensing devices to monitor the activities of daily living of the elderly population [(5), (6), (7)], however, very few are focusing on automatically detecting a fall and responding appropriately. As such, initial work on the intelligent ERS has mainly focused on the development and initial testing of the fall detection system.


Currently, the fall detection component of this new ERS uses computer vision consisting of a ceiling mounted digital camera to locate and track the occupant when they enter the room. Using simple background subtraction algorithms (8) combined with a connective-component labeling technique (5), the image is processed and the shape of the person is determined and extracted from the background as a silhouette. Various features and geometric properties of this silhouette are then calculated by the system which are used to characterize the person’s posture—i.e. depending on whether the person is standing, sitting or lying down, the silhouette will adopt different shapes and sizes. Combining information on the change in properties such as the area or perimeter of the silhouette with reduced motion of the person, and comparing these values with pre-set thresholds, the sensing agent is able to detect a fall. Once a fall has been detected, the system prompts the user to see if they are okay. It uses voice recognition software to “listen” for the user’s response, or lack thereof, upon which it decides which actions/responses to execute. For example if the person has fallen and does not respond, the system will automatically contact the closest emergency facility.


An initial pilot study was completed using four participants of varying stature and gender. Each participant performed various combinations of four different actions, namely walking, walking with arms extended, stooping and falling. Using a sample of 98 varied actions (38 falls and 60 non-falls), the system was able to reliably detect falls with an 86.8% accuracy rate with a 13.2% chance of missing a fall. Furthermore, the system triggered a false alarm in 7% of these cases. It should be noted that these preliminary results were obtained in ideal conditions and are constrained to the limitations of the system that will be further investigated.


Although the current system has shown promising results, the data collected has indicated several remaining challenges that need to be solved in order for the system to be more robust. For example, adaptive algorithms will need to be implemented that allow the system to operate reliably in a potentially cluttered and changing background. Other remaining challenges and future work also include:

Work to date has provided some evidence that using intelligent computer systems to ensure the safety of seniors in their home and monitor their daily activities seems both a practical and feasible solution. Environments and homes that can intelligently aid in caregiving could play a very significant role in enhancing aging-in-place, and thus help to reduce the burden of care on the healthcare industry.


  1. Elford, R. W. (1994). Prevention of household and recreational injuries in the elderly. Canadian Task Force on the Periodic Health Examination. Canadian Guide to Clinical Preventative Health Care. Ottawa, Health Canada,912-920.
  2. Shaw, F. E. (2003). Falls in Older People with Dementia. Geriatrics and Aging, 6(7), 37-40.
  3. Johnson, M, Cusick A, Chang, S. (2001). Home-Screen: A short scale to measure fall risk in the home. Public Health Nursing, 18(3), 169-177.
  4. Tinetti, M.E, Liu, W, Claus, E.B. (1993). Predictors and prognosis of inability to get up after falls among elderly persons. Journal of the American Medical Association, 269, 65.
  5. Wren, C., Azarbayejani, A. et al. (1997). Pfinder: Real-Time Tracking of the Human Body. IEEE Transactions on Pattern Analysis and Machine Intelligence, 19(7), 780-785.
  6. McKenna, S. J., Jabri S., et al. (2000). Tracking Groups of People. Computer Vision and Image Understanding, 80(1), 42-56.
  7. Christensen, H.I., Philips, P.J. (2002). Empirical Evaluation Methods in Computer Vision. Riveredge, NJ: World Scientific Publishing.
  8. Ronse, C, Denijver P.A. (1984). Connected Components in Binary Images: The Detection Problem. Hertfordshire, England: Research Studies Press


This research is funded in part by a NET (New Emerging Team) grant from CIHR through the Institute on Aging (grant #NET-54025).


Tracy Lee, M.A.Sc.
Intelligent Assistive Technology and Systems Laboratory, University of Toronto
500 University Ave.
Toronto, Ontario, M5G 1V7
Tel: (416) 946-8573

RESNA Conference Logo