A Two-Stage Real-Time Activity Monitoring System

Most existing human activity classification systems require a large training dataset to construct statistical models for each activity of interest. This may be impractical in many cases. In this paper, we propose a two stage classifier in order to alleviate the requirement of a large training data. In the first stage, we identify simple events such as sit, stand and walk using three triaxial accelerometers. The second stage recognizes a more complex activity using a Markov model that temporally links the events classified in the first stage. Experimental results demonstrate the feasibility of our proposed system.