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 2019-09-20 11:40
Accurate Pedestrian Detection in Thermal Images for FPGA
Ryosuke Kuramochi, Masayuki Shimoda, Youki Sada, Shimpei Sato, Hiroki Nakahara (titech) RECONF2019-26
Abstract (in Japanese) (See Japanese page) 
(in English) Since thermal cameras can detect the heat of objects, they can be used even if there is no light.
Therefore, object detection with a thermal camera is reliable and attracts much attention.
However, it is hard to obtain high detection accuracy with thermal images due to various drawbacks of them.
And also, when incorporating object detection into embedded systems such as a surveillance camera, there is a trade-off between accuracy and processing time.
To realize the real-time pedestrian detection system with high accuracy on FPGA, we propose to use a pair of a thermal image, and a background-subtracted one as an input image for YOLOv2 object detector.
We also propose a pedestrian detector system which contains a vectorizing unit.
With the vectorizing unit, we can store only valid values of a feature map together in the same memory address, resulting in calculations with a high degree of parallelism.
Compared to the conventional pedestrian detection system for thermal images, our method obtains 29 points higher F-score and operates 3.3 times faster.
From these results, our pedestrian detection system is more suitable for embedded surveillance systems.
Keyword (in Japanese) (See Japanese page) 
(in English) Deep Learning / CNN / FPGA / Thermal Image / / / /  
Reference Info. IEICE Tech. Rep., vol. 119, no. 208, RECONF2019-26, pp. 31-36, Sept. 2019.
Paper # RECONF2019-26 
Date of Issue 2019-09-12 (RECONF) 
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 RECONF2019-26

Conference Information
Committee RECONF  
Conference Date 2019-09-19 - 2019-09-20 
Place (in Japanese) (See Japanese page) 
Place (in English) KITAKYUSHU Convention Center 
Topics (in Japanese) (See Japanese page) 
Topics (in English) Reconfigurable Systems, etc. 
Paper Information
Registration To RECONF 
Conference Code 2019-09-RECONF 
Language Japanese 
Title (in Japanese) (See Japanese page) 
Sub Title (in Japanese) (See Japanese page) 
Title (in English) Accurate Pedestrian Detection in Thermal Images for FPGA 
Sub Title (in English)  
Keyword(1) Deep Learning  
Keyword(2) CNN  
Keyword(3) FPGA  
Keyword(4) Thermal Image  
Keyword(5)  
Keyword(6)  
Keyword(7)  
Keyword(8)  
1st Author's Name Ryosuke Kuramochi  
1st Author's Affiliation Tokyo Institute of Technology (titech)
2nd Author's Name Masayuki Shimoda  
2nd Author's Affiliation Tokyo Institute of Technology (titech)
3rd Author's Name Youki Sada  
3rd Author's Affiliation Tokyo Institute of Technology (titech)
4th Author's Name Shimpei Sato  
4th Author's Affiliation Tokyo Institute of Technology (titech)
5th Author's Name Hiroki Nakahara  
5th Author's Affiliation Tokyo Institute of Technology (titech)
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 2019-09-20 11:40:00 
Presentation Time 20 minutes 
Registration for RECONF 
Paper # RECONF2019-26 
Volume (vol) vol.119 
Number (no) no.208 
Page pp.31-36 
#Pages
Date of Issue 2019-09-12 (RECONF) 


[Return to Top Page]

[Return to IEICE Web Page]


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