Efficient Retrieval of Human Motion Episodes Based on Indexed
Motion-Word Representations

J 2021

Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations

BUDÍKOVÁ, Petra, Jan SEDMIDUBSKÝ, Ján HORVÁTH and Pavel ZEZULA

Basic information

Original name

Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations

Authors

BUDÍKOVÁ, Petra (203 Czech Republic, belonging to the institution), Jan SEDMIDUBSKÝ (203 Czech Republic, belonging to the institution), Ján HORVÁTH (703 Slovakia, belonging to the institution) and Pavel ZEZULA (203 Czech Republic, belonging to the institution)

Edition

International Journal of Semantic Computing, World Scientific Publishing, 2021, 1793-351X

Other information

Language

English

Type of outcome

Článek v odborném periodiku

Field of Study

10200 1.2 Computer and information sciences

Country of publisher

Singapore

Confidentiality degree

není předmětem státního či obchodního tajemství

References:

URL

RIV identification code

RIV/00216224:14330/21:00118932

Organization unit

Faculty of Informatics

DOI

http://dx.doi.org/10.1142/S1793351X21400031

UT WoS

000670288200004

Keywords in English

human motion data; motion episodes; text-based processing; indexing

Abstract

V originále

With the increasing availability of human motion data captured in the form of 2D or 3D skeleton sequences, more complex motion recordings need to be processed. In this paper, we focus on similarity-based indexing and efficient retrieval of motion episodes - medium-sized skeleton sequences that consist of multiple semantic actions and correspond to some logical motion unit (e.g., a figure skating performance). As a first step towards efficient retrieval, we apply the motion-word technique to transform spatio-temporal skeleton sequences into compact text-like documents. Based on these documents, we introduce a two-phase retrieval scheme that first finds a set of candidate query results and then re-ranks these candidates with more expensive application-specific methods. We further index the motion-word documents using inverted files, which allows us to retrieve the candidate documents in an efficient and scalable manner. We also propose additional query-reduction techniques that accelerate both the retrieval phases by removing semantically irrelevant parts of the motion query. Experimental evaluation is used to analyze the effects of the individual proposed techniques of the retrieval efficiency and effectiveness.

Links

GA19-02033S, research and development project

Name: Vyhledávání, analytika a anotace datových toků lidských pohybů

Investor: Czech Science Foundation

Citovat

BUDÍKOVÁ, Petra, Jan SEDMIDUBSKÝ, Ján HORVÁTH and Pavel ZEZULA. Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations. International Journal of Semantic Computing. World Scientific Publishing, 2021, vol. 15, No 2, p. 189-213. ISSN 1793-351X. Available from: https://dx.doi.org/10.1142/S1793351X21400031.

@article{1760617,
   author = {Budíková, Petra and Sedmidubský, Jan and Horváth, Ján and Zezula, Pavel},
   article_number = {2},
   doi = {http://dx.doi.org/10.1142/S1793351X21400031},
   keywords = {human motion data; motion episodes; text-based processing; indexing},
   language = {eng},
   issn = {1793-351X},
   journal = {International Journal of Semantic Computing},
   title = {Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations},
   url = {https://dx.doi.org/10.1142/S1793351X21400031},
   volume = {15},
   year = {2021}
}

TY  - JOUR
ID  - 1760617
AU  - Budíková, Petra - Sedmidubský, Jan - Horváth, Ján - Zezula, Pavel
PY  - 2021
TI  - Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations
JF  - International Journal of Semantic Computing
VL  - 15
IS  - 2
SP  - 189-213
EP  - 189-213
PB  - World Scientific Publishing
SN  - 1793351X
KW  - human motion data
KW  - motion episodes
KW  - text-based processing
KW  - indexing
UR  - https://dx.doi.org/10.1142/S1793351X21400031
N2  - With the increasing availability of human motion data captured in the form of 2D or 3D skeleton sequences, more complex motion recordings need to be processed. In this paper, we focus on similarity-based indexing and efficient retrieval of motion episodes - medium-sized skeleton sequences that consist of multiple semantic actions and correspond to some logical motion unit (e.g., a figure skating performance). As a first step towards efficient retrieval, we apply the motion-word technique to transform spatio-temporal skeleton sequences into compact text-like documents. Based on these documents, we introduce a two-phase retrieval scheme that first finds a set of candidate query results and then re-ranks these candidates with more expensive application-specific methods. We further index the motion-word documents using inverted files, which allows us to retrieve the candidate documents in an efficient and scalable manner. We also propose additional query-reduction techniques that accelerate both the retrieval phases by removing semantically irrelevant parts of the motion query. Experimental evaluation is used to analyze the effects of the individual proposed techniques of the retrieval efficiency and effectiveness.
ER  -

BUDÍKOVÁ, Petra, Jan SEDMIDUBSKÝ, Ján HORVÁTH and Pavel ZEZULA. Efficient Retrieval of Human Motion Episodes Based on Indexed Motion-Word Representations. \textit{International Journal of Semantic Computing}. World Scientific Publishing, 2021, vol.~15, No~2, p.~189-213. ISSN~1793-351X. Available from: https://dx.doi.org/10.1142/S1793351X21400031.

Detailed Information on Publication Record