Abstract—This research developed a prototype to recognize the activities of people with social interaction and communication impairments using two commercial grade infrared motion sensing input devices. The model uses skeletal joints in the form of (x, y, z) coordinates to perform computationally inexpensive and efficient real time indoor monitoring. The prototype works well for a room with the size of 15 ft. by 15 ft. where it can detect two human subjects simultaneously. The prototype promotes non-invasive monitoring, no wearable sensor is required on the person under monitoring and no identifiable face images are stored. The evaluation shows that it is able to recognize the accident-critical activities, self-injurious activities and inactivity with satisfactory recognition and false alarm rates.
 

Index Terms—activity recognition, human body joints, gesture, accident-critical, self-injurious, inactivity, fall detection, people with disabilities, motion sensor

Cite: Theng B. Lau, Ann C. Ong, and Fernando A. Putra, "Non-Invasive Monitoring of People with Disabilities via Motion Detection," International Journal of Signal Processing Systems, Vol. 2, No. 1, pp. 37-41, June 2014. doi: 10.12720/ijsps.2.1.37-41