IEICE Technical Committee Submission System
Conference Paper's Information
Online Proceedings
[Sign in]
Tech. Rep. Archives
 Go Top Page Go Previous   [Japanese] / [English] 

Paper Abstract and Keywords
Presentation 2017-02-18 15:00
Evaluation of Triple-Stream Convolutional Network for Action Recognition
Dichao Liu, Yu Wang, Jien Kato, Kenji Mase (Nagoya Univ.) PRMU2016-168 CNR2016-35
Abstract (in Japanese) (See Japanese page) 
(in English) Recently, Two-Stream Convolutional Network has achieved remarkable performance. For example, by using static frames and optical flows as inputs and training the fused networks together, local motion information can be explicitly captured, bringing about noticeable improvement for end-to-end performance.

On the other hand, dynamic image, whose pixel values are rescaled from parameters of a ranking machine encoding the frames’ temporal variety, have also been confirmed to provide complimentary information to spatial appearance. Inspired by these works, we proposed Triple-Stream Convolutional Network by adding a third stream of network whose input is dynamic image. In this paper, we implemented the proposed Triple-Stream Convolutional Network, and evaluated it in two aspects: (1) how the overall end-to-end classification performance can be benefited by adding the third stream; (2) which way is efficient to use the trained Triple-Stream Convolutional Network in classification.
Keyword (in Japanese) (See Japanese page) 
(in English) Action recognition / ConvNets / Deep Learning / Fusion / / / /  
Reference Info. IEICE Tech. Rep., vol. 116, no. 461, PRMU2016-168, pp. 91-94, Feb. 2017.
Paper # PRMU2016-168 
Date of Issue 2017-02-11 (PRMU, CNR) 
ISSN Print edition: ISSN 0913-5685    Online edition: ISSN 2432-6380
Copyright
and
reproduction
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034)
Download PDF PRMU2016-168 CNR2016-35

Conference Information
Committee PRMU CNR  
Conference Date 2017-02-18 - 2017-02-19 
Place (in Japanese) (See Japanese page) 
Place (in English)  
Topics (in Japanese) (See Japanese page) 
Topics (in English)  
Paper Information
Registration To PRMU 
Conference Code 2017-02-PRMU-CNR 
Language English 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Evaluation of Triple-Stream Convolutional Network for Action Recognition 
Sub Title (in English)  
Keyword(1) Action recognition  
Keyword(2) ConvNets  
Keyword(3) Deep Learning  
Keyword(4) Fusion  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Dichao Liu  
1st Author's Affiliation Nagoya University (Nagoya Univ.)
2nd Author's Name Yu Wang  
2nd Author's Affiliation Nagoya University (Nagoya Univ.)
3rd Author's Name Jien Kato  
3rd Author's Affiliation Nagoya University (Nagoya Univ.)
4th Author's Name Kenji Mase  
4th Author's Affiliation Nagoya University (Nagoya Univ.)
5th Author's Name  
5th Author's Affiliation ()
6th Author's Name  
6th Author's Affiliation ()
7th Author's Name  
7th Author's Affiliation ()
8th Author's Name  
8th Author's Affiliation ()
9th Author's Name  
9th Author's Affiliation ()
10th Author's Name  
10th Author's Affiliation ()
11th Author's Name  
11th Author's Affiliation ()
12th Author's Name  
12th Author's Affiliation ()
13th Author's Name  
13th Author's Affiliation ()
14th Author's Name  
14th Author's Affiliation ()
15th Author's Name  
15th Author's Affiliation ()
16th Author's Name  
16th Author's Affiliation ()
17th Author's Name  
17th Author's Affiliation ()
18th Author's Name  
18th Author's Affiliation ()
19th Author's Name  
19th Author's Affiliation ()
20th Author's Name  
20th Author's Affiliation ()
Speaker Author-1 
Date Time 2017-02-18 15:00:00 
Presentation Time 25 minutes 
Registration for PRMU 
Paper # PRMU2016-168, CNR2016-35 
Volume (vol) vol.116 
Number (no) no.461(PRMU), no.462(CNR) 
Page pp.91-94 
#Pages
Date of Issue 2017-02-11 (PRMU, CNR) 


[Return to Top Page]

[Return to IEICE Web Page]


The Institute of Electronics, Information and Communication Engineers (IEICE), Japan