Activity classification using mobile phones is useful for identifying training activities, then capturing short periods of high frequency training data and capturing and archiving appropriate training statistics for various training activities. Some available smart phone training information systems classify the negative case of resting during a training session but none actively detect training activities and classify type of activity. It is widely perceived that activity classification is useful but few activity classifiers are available for smart phones. We test one activity classifier for the Android platform that is able to run as a background application without an obvious impact on battery life and which reported high levels of accuracy. The reported accuracy was not achieved during testing, in part because users were applying different criteria to determine accuracy than developers. A smart phone classifier was developed adding several techniques to increase usefulness and accuracy as perceived by users. These included detecting device states where inferring user activity was not possible, limiting the range of activities to those that can be reliably detected, eliminating dependence on device orientation, presenting aggregated information graphically and web based archiving of activity history. The classifier can be used for detecting levels of exercise undertaken, detecting occurrence of training activities and for messaging other applications to trigger collection of appropriate detailed information and summary statistics. Combining activity information with applications inferring lifestyle activities from location based data would enhance the usefulness of both applications.
© Copyright 2011 Procedia Engineering. Elsevier. Kaikki oikeudet pidätetään. / Kääntänyt https://www.DeepL.com/Translator