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 2022-03-11 14:30
Background Mixup Data Augmentation for Hand and Object-in-Contact Detection
Koya Tango, Takehiko Ohkawa, Ryosuke Furuta, Yoichi Sato (UTokyo) PRMU2021-82
Abstract (in Japanese) (See Japanese page) 
(in English) Detecting the position of human hands and an object-in-contact from an image is vital for understanding a user’s actions in a video and logging his or her daily activities.
For training an object detector, Mixup, which overlays two training images to mitigate data bias, has been studied as an empirically powerful augmentation.
However, in our task of hand and object-in-contact detection, mixing two human activity images produces an unintended bias in the augmented training data, e.g., the concentration of hands and objects in a specific region degrades the ability of the hand and object-in-contact detector to identify object boundaries.
In this work, we propose Background Mixup data augmentation that leverages data mixing regularization while reducing the unintended effect in hand and object-in-contact detection.
Specifically, instead of mixing two images where a hand and object-in-contact pair appears, we mix a target training image with background images extracted from external image sources and use them to train the detector.
Our experimental study demonstrated that the proposed method effectively reduced false positives and improved the performance of hand and object-in-contact detection in supervised and semi-supervised learning settings.
Keyword (in Japanese) (See Japanese page) 
(in English) Data Augmentation / Hand and Object-in-Contact Detection / Mixup / / / / /  
Reference Info. IEICE Tech. Rep., vol. 121, no. 427, PRMU2021-82, pp. 139-144, March 2022.
Paper # PRMU2021-82 
Date of Issue 2022-03-03 (PRMU) 
ISSN 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 PRMU2021-82

Conference Information
Committee PRMU IPSJ-CVIM  
Conference Date 2022-03-10 - 2022-03-11 
Place (in Japanese) (See Japanese page) 
Place (in English) Online 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Differentiable rendering 
Paper Information
Registration To PRMU 
Conference Code 2022-03-PRMU-CVIM 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Background Mixup Data Augmentation for Hand and Object-in-Contact Detection 
Sub Title (in English)  
Keyword(1) Data Augmentation  
Keyword(2) Hand and Object-in-Contact Detection  
Keyword(3) Mixup  
Keyword(4)  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Koya Tango  
1st Author's Affiliation The University of Tokyo (UTokyo)
2nd Author's Name Takehiko Ohkawa  
2nd Author's Affiliation The University of Tokyo (UTokyo)
3rd Author's Name Ryosuke Furuta  
3rd Author's Affiliation The University of Tokyo (UTokyo)
4th Author's Name Yoichi Sato  
4th Author's Affiliation The University of Tokyo (UTokyo)
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 2022-03-11 14:30:00 
Presentation Time 15 minutes 
Registration for PRMU 
Paper # PRMU2021-82 
Volume (vol) vol.121 
Number (no) no.427 
Page pp.139-144 
#Pages
Date of Issue 2022-03-03 (PRMU) 


[Return to Top Page]

[Return to IEICE Web Page]


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