Paper Abstract and Keywords |
Presentation |
2013-02-21 16:10
[Poster Presentation]
Evaluation and Expansion of Recognition methods Using Unlabeled Regions by means of Self-training Daichi Tanabe, Toru Tamaki, Bisser Raytchev, Kazufumi Kaneda, Tetsushi Koide, Yoko Kominami, Rie Miyaki, Taiji Matsuo, Shigeto Yoshida, Shinji Tanaka (Hiroshima Univ.) PRMU2012-147 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this paper, we report the results of the problems discussed for the case in which the recognition of NBI endoscopic image,using the unlabeled region by means of self-training.
Takeda et al have proposed a method creating a large database using the Self-training in order increase images for learning.
But, this method is no reliable because it uses an image that has not received confirmation of doctor.
So in this study, we consider the conventional method and experiment with a method improving its problem by using a larger data set.
As a result, we show the properties of the conventional method. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Colorectal Tumor / Unlabeled region / NBI Image / Self-training / Active-learning / / / |
Reference Info. |
IEICE Tech. Rep., vol. 112, no. 441, PRMU2012-147, pp. 103-104, Feb. 2013. |
Paper # |
PRMU2012-147 |
Date of Issue |
2013-02-14 (PRMU) |
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 |
PRMU2012-147 |
|