Paper Abstract and Keywords |
Presentation |
2023-11-16 16:50
On Food Plant Classification from Luehdorfia Japonica Images using Multi-label Classification ABN Tsubasa Hirakawa, Takaaki Arai, Takayoshi Yamashita, Hironobu Fujiyoshi, Yuichi Oba, Hiromichi Fukui (Chubu Univ.), Masaya Yago (Tokyo Univ.) PRMU2023-27 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Butterfly is a familiar taxon. Because of the abundance of specimens and the ease of comparison between specimens, regional variation in butterfly spots is well known, which is related to various factors such as geological history, topography, climate, and food plants. In this study, we focus on the relationship between regional variation in butterfly spots and the distribution of food plants, and aim to clarify the relationship by classifying food plants based on images of butterfly spots. Specifically, we will focus on Luehdorfia japonica, a species of butterfly with known geographic variation in butterfly spots and a highly understood distribution. We will create a spotted butterfly image classification dataset based on digital specimens and collection site metadata. We show that the multi-label Attention Branch Netowrk can be trained on the dataset to accurately classify the food plants from the butterfly images, and that the analysis of the attention map provides the same basis for decision-making as the expert knowledge. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Attention Branch Network / Multi-label Classification / Visual Explanation / Luehdorfia japonica / Local Variation / / / |
Reference Info. |
IEICE Tech. Rep., vol. 123, no. 266, PRMU2023-27, pp. 62-67, Nov. 2023. |
Paper # |
PRMU2023-27 |
Date of Issue |
2023-11-09 (PRMU) |
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 |
PRMU2023-27 |
|