Paper Abstract and Keywords |
Presentation |
2018-03-03 14:25
Study of Visual Cognition Estimation based on Gaze Movements and target Visual Cognition Difficultly Hideho Sakaguchi (NAIST/ATR), Akira Utsumi (ATR), Kenji Susami (ATR/Kindai Univ.), Tadahisa Kondo (Kogakuin Univ.), Masayuki Kanbara (NAIST/ATR), Norihiro Hagita (ATR/NAIST) HIP2017-96 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Proper estimation of human visual awareness is significant to realize an efficient support for visual tasks such as vehicle driving. We examine a method to estimate occurrences of human visual cognition. Several methods based on gaze movement have been proposed to estimate human visual awareness where simple thresholding rules on target-gaze angles are employed. However, those methods take no account of the variety of target visual appearances. In our previous studies, we clarified the relationship between gaze behavior in visual search tasks and the target’s VCDs (Visual Cognition Difficulties) defined as performance values on visual search tasks for a particular target. In this paper, we evaluate the estimation performance by using machine-learning based implementation. The results suggest the improvement of performance for visual cognition estimation by considering the target’s VCDs. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
gaze movements / Visual Cognition Difficultly / machine learning / Convolutional Neural Network (CNN) / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 117, no. 470, HIP2017-96, pp. 17-20, March 2018. |
Paper # |
HIP2017-96 |
Date of Issue |
2018-02-24 (HIP) |
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 |
HIP2017-96 |
|