Physical activities detection plays a vital role to healthcare professionals who would like to monitor patients remotely and to develop context-sensitive systems. Major number of physical activity detection systems use accelerometers to collect data from different parts of the body. Since those approaches have limitations from users' point of view, we have used smart phones that are coming with built-in accelerometers and gyroscopes. We have proposed and developed three novel approaches for activity recognition. Firstly, we have developed a multimodal system where we used pressure sensor data from shoes along with accelerometers and gyroscope data from smart phone. Again, we have presented the details of our novel activity detection system along with evaluation. In the second approach, we considered our sensor data as time series shapelets and apply recently developed algorithms to differentiate those shapelets. Finally, we applied Gaussian Mixture Models with time-delay embedding for detecting different activities.
COMPSAC 2015Kawsar, Ferdaus, Md Kamrul Hasan, Richard Love, and Sheikh Iqbal Ahamed. "A novel activity detection system using plantar pressure sensors and smartphone." In Computer Software and Applications Conference (COMPSAC), 2015 IEEE 39th Annual, vol. 1, pp. 44-49. IEEE, 2015.
ICOST 2016Kawsar, Ferdaus, Md Kamrul Hasan, Tanvir Roushan, Sheikh Iqbal Ahamed, William C. Chu, and Richard Love. "Activity Detection Using Time-Delay Embedding in Multi-modal Sensor System." In International Conference on Smart Homes and Health Telematics, pp. 489-499. Springer, Cham, 2016.