Paper Abstract and Keywords |
Presentation |
2020-09-09 14:00
Shitsukan Representation Based on Kansei Model Using Neural Style Feature Natsuki Sunda, Iori Tani (Kwansei Gakuin Univ.), Kensuke Tobitani (The Univ. of Nagasaki), Atsushi Takemoto, Yusuke Tani, Noriko Nagata (Kwansei Gakuin Univ.), Nobufumi Morita (Couture Digital) MVE2020-18 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In this research, we focus on affective texture, which comprises the visual impressions evoked by surface properties, such as fine patterns and roughness, and propose a method for texture synthesis with desired affective texture. First, we modeled the relationships using a lasso regression between affective texture quantified by psychological experiments and style features which are texture features extracted from CNN. After that, based on the obtained models, we performed texture synthesis by manipulating style features that contribute to the affective texture. As a result, the generation results matched human intuition, confirming the effectiveness of our method. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Affective Texture / Texture Synthesis / Fashion / CNN / Style Transfer / Lasso Regression / / |
Reference Info. |
IEICE Tech. Rep., vol. 120, no. 160, MVE2020-18, pp. 38-43, Sept. 2020. |
Paper # |
MVE2020-18 |
Date of Issue |
2020-09-01 (MVE) |
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 |
MVE2020-18 |
|