Paper Abstract and Keywords |
Presentation |
2020-03-05 14:20
A study on image captioning considering its imageability Kazuki Umemura, Marc Aurel Kastner, Ichiro Ide, Yasutomo Kawanishi, Takatsugu Hirayama (Nagoya Univ.), Keisuke Doman (Chukyo Univ.), Daisuke Deguchi, Hiroshi Murase (Nagoya Univ.) IMQ2019-48 IE2019-130 MVE2019-69 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
We propose an imageability-aware image captioning method tailoring generated captions to various applications. In this study, we first extend an existing image captioning dataset by augmenting its captions. Then, an imageability score for each augmented caption is calculated. A modified image captioning model is trained using this extended dataset to generate captions tailored to a specified imageability score. The evaluation shows the possibility that the extended dataset and the proposed method can generate imageability-aware captions. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Multimedia proccesing / image captioning / psycholinguistics / semantic gap / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 119, no. 457, MVE2019-69, pp. 165-169, March 2020. |
Paper # |
MVE2019-69 |
Date of Issue |
2020-02-27 (IMQ, IE, MVE) |
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 |
IMQ2019-48 IE2019-130 MVE2019-69 |
|