Classification Method for Human Locomotion
Ramona Luca, Silviu-Ioan Bejinariu · 2018
A framework to integrate knowledge representation in human locomotion is proposed. A descriptive statistics was made on a set of motion parameters which were extracted from video collections and stored in a relational database. The k-means clustering algorithm and the k-nearest neighbors classifier were applied in the space of the statistical parameters and a comparison between clustering ratio and classification ratio was made. The experiments were performed using the KTH and the Weizmann video datasets which describe human locomotion. The parameters extracted from Weizmann video collection were used to validate the proposed method. The main purpose of this research is to achieve an ontology with clear and accurate rules that describe human locomotion.