研究者情報 | 神奈川大学

イシイ　ノブアキ Ishii Nobuaki 石井　信明所属神奈川大学工学部経営工学科神奈川大学大学院工学研究科工学専攻（経営工学領域）職種教授
言語種別	英語
発行・発表の年月	2003/07
形態種別	著書
査読	査読あり
標題	Crack Detection on Brick Walls by Convolutional Neural Networks Using the Methods of Sub-dataset Generation and Matching
執筆形態	共著
掲載誌名	Deep Learning Theory and Applications
掲載区分	国外
出版社・発行元	Springer
巻・号・頁	pp.134-150
総ページ数	17
担当区分	最終著者
国際共著	国際共著
著者・共著者	Talukder, M. H., Ota, S., Takanokura, M., and Ishii, N.
概要	In this study, sub-dataset generation and matching methods are proposed to improve the performance of crack detection in brick walls using CNN. CNN training is conducted with each sub-dataset generated by the proposed sub-dataset generation method, while crack detection is performed using a properly trained CNN that is selected using the proposed matching method. For numerical experiments, training datasets are first prepared by manual image cropping and rotation, after which the performance of crack detection is evaluated by cross-validation. Numerical experiments show that the proposed method improves crack detection in brick walls. This study will help to ensure the safety of structures as well as the safety of human life.
DOI	https://doi.org/10.1007/978-3-031-37320-6
ISBN	978-3-031-37319-0
ISSN	978-3-031-37319-0