Enhancing Effectiveness of Descriptors for Searching and
Recognition in Motion Capture Data

D 2017

Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data

SEDMIDUBSKÝ, Jan, Petr ELIÁŠ and Pavel ZEZULA

Basic information

Original name

Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data

Authors

SEDMIDUBSKÝ, Jan (203 Czech Republic, guarantor, belonging to the institution), Petr ELIÁŠ (203 Czech Republic, belonging to the institution) and Pavel ZEZULA (203 Czech Republic, belonging to the institution)

Edition

Neuveden, 19th IEEE International Symposium on Multimedia, p. 240-243, 4 pp. 2017

Publisher

IEEE Computer Society

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10201 Computer sciences, information science, bioinformatics

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

storage medium (CD, DVD, flash disk)

RIV identification code

RIV/00216224:14330/17:00094977

Organization unit

Faculty of Informatics

ISBN

978-1-5386-2937-6

DOI

http://dx.doi.org/10.1109/ISM.2017.39

UT WoS

000454605200033

Keywords in English

motion capture data; similarity-based comparison; motion images; joint weights; deep convolutional neural network; distance function; action recognition

Abstract

V originále

Computer-aided analyses of motion capture data require an effective and efficient concept of motion similarity. Traditional methods generally compare motion sequences by applying time-warping techniques to high-dimensional trajectories of joints. An increasing effectiveness of machine-learning techniques, such as deep convolutional neural networks, brings new possibilities for similarity comparison. Inspired by recent advances in neural networks and image processing, we propose new variants of transformation of motion sequences into 2D images. The generated images are used to fine-tune a neural network from which 4,096D features are extracted and compared by a modified Euclidean distance. The proposed concept is not only efficient but also very effective and outperforms existing methods on a challenging dataset with 130 categories.

Links

GBP103/12/G084, research and development project

Name: Centrum pro multi-modální interpretaci dat velkého rozsahu

Investor: Czech Science Foundation

Citovat

SEDMIDUBSKÝ, Jan, Petr ELIÁŠ and Pavel ZEZULA. Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data. In 19th IEEE International Symposium on Multimedia. Neuveden: IEEE Computer Society, 2017, p. 240-243. ISBN 978-1-5386-2937-6. Available from: https://dx.doi.org/10.1109/ISM.2017.39.

@inproceedings{1390238,
   author = {Sedmidubský, Jan and Eliáš, Petr and Zezula, Pavel},
   address = {Neuveden},
   booktitle = {19th IEEE International Symposium on Multimedia},
   doi = {http://dx.doi.org/10.1109/ISM.2017.39},
   keywords = {motion capture data; similarity-based comparison; motion images; joint weights; deep convolutional neural network; distance function; action recognition},
   howpublished = {paměťový nosič},
   language = {eng},
   location = {Neuveden},
   isbn = {978-1-5386-2937-6},
   pages = {240-243},
   publisher = {IEEE Computer Society},
   title = {Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data},
   year = {2017}
}

TY  - JOUR
ID  - 1390238
AU  - Sedmidubský, Jan - Eliáš, Petr - Zezula, Pavel
PY  - 2017
TI  - Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data
PB  - IEEE Computer Society
CY  - Neuveden
SN  - 9781538629376
KW  - motion capture data
KW  - similarity-based comparison
KW  - motion images
KW  - joint weights
KW  - deep convolutional neural network
KW  - distance function
KW  - action recognition
N2  - Computer-aided analyses of motion capture data require an effective and efficient concept of motion similarity. Traditional methods generally compare motion sequences by applying time-warping techniques to high-dimensional trajectories of joints. An increasing effectiveness of machine-learning techniques, such as deep convolutional neural networks, brings new possibilities for similarity comparison. Inspired by recent advances in neural networks and image processing, we propose new variants of transformation of motion sequences into 2D images. The generated images are used to fine-tune a neural network from which 4,096D features are extracted and compared by a modified Euclidean distance. The proposed concept is not only efficient but also very effective and outperforms existing methods on a challenging dataset with 130 categories.
ER  -

SEDMIDUBSKÝ, Jan, Petr ELIÁŠ and Pavel ZEZULA. Enhancing Effectiveness of Descriptors for Searching and Recognition in Motion Capture Data. In \textit{19th IEEE International Symposium on Multimedia}. Neuveden: IEEE Computer Society, 2017, p.~240-243. ISBN~978-1-5386-2937-6. Available from: https://dx.doi.org/10.1109/ISM.2017.39.

Detailed Information on Publication Record