Tree-Based Classifiers on Human Locomotion Statistical Parameters

Abstract

In order to describe the human locomotion types based on the extracted parameters from video sequences, data mining techniques are used to discover classification rules and generating decision trees. Applying J48, JRIP and Random forest classifiers in the space of extracted statistical parameters from video sequences, in the evaluation process, the best result was obtained by the Random forest classifier. The purpose of this research is to represent knowledge obtained from video sequences about human locomotion, in form of ontology, with rules that describe the walking, jogging and running locomotion types.