Towards Scalable Retrieval of Human Motion Episodes

Investor logo

Warning

This publication doesn't include Faculty of Arts. It includes Faculty of Informatics. Official publication website can be found on muni.cz.
Authors

BUDÍKOVÁ Petra SEDMIDUBSKÝ Jan HORVÁTH Ján ZEZULA Pavel

Year of publication 2020
Type Article in Proceedings
Conference 22nd IEEE International Symposium on Multimedia (ISM)
MU Faculty or unit

Faculty of Informatics

Citation
Doi http://dx.doi.org/10.1109/ISM.2020.00015
Keywords skeleton data; human motion retrieval; motion episodes; text-based processing
Description With the increasing availability of human motion data captured in the form of 2D/3D skeleton sequences, more complex motion recordings need to be processed. In this paper, we study the problem of similarity-based matching of medium-sized unsegmented skeleton sequences, which we denote as motion episodes. We first apply standard pose-based approaches for matching episodes and analyze their shortcomings. Then, we adopt a recent segment-based approach that transforms episode data into a text-like representation, and apply mature text-processing techniques for matching episodes. We demonstrate that this text-based approach achieves promising results in the terms of both effectiveness and efficiency, and can be further indexed to implement scalable episode retrieval.
Related projects:

You are running an old browser version. We recommend updating your browser to its latest version.