Retrieving similar movements in motion capture data

Abstract

Efficient and effective similarity retrieval in large human motion databases is valuable in many fields such as computer animation, security research, sports, and medicine. This demonstration paper presents a content-based retrieval system that is able to examine database motions and locate all their sub-motions similar to a query motion example. The proposed system does not require any textual annotations nor explicit knowledge of the data and can deal with spatio-temporal variances of individual human motions. The system is presented as an online web application indexing a real-life 68-minute human motion database. © 2013 Springer-Verlag.