A gait analysis technique to model user presences in an office scenario is presented in this chapter. In contrast with other approaches, we use unobtrusive sensors, i.e., an array of Kinect devices, to detect some features of interest. In particular, the position and the spatio-temporal evolution of some skeletal joints are used to define a set of gait features, which can be either static (e.g., person height) or dynamic (e.g., gait cycle duration). Data captured by multiple Kinects is merged to detect dynamic features in a longer walk sequence. The approach proposed here was been evaluated by using three classifiers (SVM, KNN, Naive Bayes) on different feature subsets.
|Titolo della pubblicazione ospite||Advances onto the Internet of Things
How Ontologies Make the Internet of Things Meaningful|
|Numero di pagine||11|
|Stato di pubblicazione||Published - 2014|
|Nome||ADVANCES IN INTELLIGENT SYSTEMS AND COMPUTING|