Authors: Lau Bee Theng, Ong Chin Ann and Hamid Bagha
This research proposed a model to recognize the common activities of people with special needs using soft computing and infrared depth sensor. We developed a rotatable single sensor which resolves the blind spots that are occluded by non-rotatable single depth sensor; it also saves on resources as one rotatable sensor can provide the view field of two sensors. The model provides computationally inexpensive and efficient algorithms to recognize the activities for real time indoor monitoring of people with disabilities. Besides, the model works in real time, non-intrusive and intuitive manner. There is no wearable device required on the person under monitoring and he or she does not have to coordinate any interface element. The prototype evaluation shows that the model is able to recognize the common activities and critical incidents effectively. It has the potential to be adapted and enhanced for monitoring people with disabilities.
Keywords: activity recognition, gesture recognition, critical incident recognition, fall detection, people with disabilities, depth sensor