Unsupervised Activity Recognition Based On Smartphone Accelerometers

With the development and popularization of smartphones, it is playing a more and more important role in people’s daily life. It can be said with certainty that smartphones have been an indispensable part of our life. With the characteristics of powerful and easy to be accepted, smartphones have been paid more and more attention by researchers. There are more and more researches based on smartphones. Smartphones equipped with accelerometers give a promising way for researchers to study on human activity recognition. And the researches in terms of health care and physical activity recognition of human activity recognition based on smartphones have gained some achievements.Three data sets in two types of daily activity and sports are used in our experiment. And our experiment is designed according to the basic process of human activity recognition. In data collection, an application based on Android system is developed. And some volunteers are invited by us to get the real physical information. In feature extraction, 19 features are selected according to our experiments. In terms of clustering algorithm, the MCODE clustering method is used by us which is applied in PPI biological clustering originally.Experimental results show that, on the daily activity recognition, our method is better than other commonly used clustering methods. The recognition rate of race walking recognition is 88%. And the recognition rate of basketball sports is 81.7%. The results show that it is viable to recognize physical activities even for complex sports using smartphone accelerometers.