Human activity recognition based on triaxial accelerometer

Activity recognition has become one of the most popular research topics. In this paper, efforts on activity recognition based on the WIMU (Wireless Inertial Measurement Unit) sensors are reported. We use triaxial accelerometer to capture the acceleration of the human body and train the models for different activities, and then employ a further application of recognizing the human activities with these models. Different settings of the sensors are compared and the importance of different features is discussed. In addition, also a real-time detection system is designed and realized, which can be used as the real-time recognition of human activities and is of great importance for the application of WIMU in the healthcare for the elderly.