window.dataLayer = window.dataLayer || []; function gtag(){dataLayer.push(arguments);} gtag('js', new Date()); gtag('config', 'UA-128570374-1');
Carlos Gutiérrez López de la Franca, Ramón Hervás, Esperanza Johnson, Tania Mondéjar, José Bravo
Journal of Ambient Intelligence and Humanized Computing, 8(4) pp. 531-549 (ISI-JCR IF=1.588 Q3). March 2017
Publication year: 2017

Machine learning methods have been proven to be a useful and powerful mechanism in classification problems and pattern recognition. It is possible to classify into different categories on the basis of a training set of data through a discrete number of characteristics and without knowing the whole intrinsic information. For that, those methods are widely extended to address Activity Recognition. One of the limitations of traditional machine learning methods is the need of a large training dataset to get an effective model with high accuracy levels. This issue is the main motivation of the present work that proposes the named Extended Body-Angles Algorithm, a method able to recognize movements through single samples of modelled movements using Kinect. Their functioning is based on the Body-Angles Algorithm. This manuscript deepens into the detection of postures (Body-Angles Algorithm) that has been generalized to movements (Extended Body-Angles Algorithm), describing also the results of the algorithm evaluation.

Keywords:Activity recognition Kinect Ubiquitous computing Behaviour-aware computing Ambient intelligence