2014 | OriginalPaper | Buchkapitel
Detecting Accelerometer Placement to Improve Activity Classification
verfasst von : Ian Cleland, Chris D. Nugent, Dewar D. Finlay, Roger Armitage
Erschienen in: XIII Mediterranean Conference on Medical and Biological Engineering and Computing 2013
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
This paper describes a method to improve the classification of everyday activities through detection of the location of an accelerometer device on the body. The detection of the device location allows an activity classification model, produced using a C4.5 decision tree and specifically tailored for that location, to be applied. Eight male subjects participated within the study. Participants wore six tri-axial accelerometers, positioned at various locations, whilst performing a number of everyday activities. A C4.5 decision tree was also used to detect the location of the accelerometer on the body which achieved an F-measure of 0.63. Based on this approach and applying the appropriate activity recognition model for the detected location improved activity recognition performance from an F-measure of 0.36 to 0.62, for the worst case, when using an activity model trained only one location.