ログイン
言語:

WEKO3

  • トップ
  • ランキング
To
lat lon distance
To

Field does not validate



インデックスリンク

インデックスツリー

メールアドレスを入力してください。

WEKO

One fine body…

WEKO

One fine body…

アイテム

  1. 40 大学院工学研究科・工学部
  2. 40D 学位論文
  3. 修士論文
  4. 2017年度

Extraction and Recognition of Shoe Logos with a Wide Variety of Appearance using Two-Stage Classifiers

http://hdl.handle.net/10076/00017875
http://hdl.handle.net/10076/00017875
68b6d54d-12af-4266-90e9-c395a1417fa1
名前 / ファイル ライセンス アクション
2017ME183.pdf 2017ME183 (18.8 MB)
Item type 学位論文 / Thesis or Dissertation(1)
公開日 2018-10-24
タイトル
タイトル Extraction and Recognition of Shoe Logos with a Wide Variety of Appearance using Two-Stage Classifiers
言語 ja
言語
言語 eng
資源タイプ
資源タイプ識別子 http://purl.org/coar/resource_type/c_46ec
資源タイプ thesis
著者 アオキ, カズノリ

× アオキ, カズノリ

en Aoki, Kazunori

ja アオキ, カズノリ

Search repository
抄録
内容記述タイプ Abstract
内容記述 A logo is a symbolic presentation that is designed not only to identify a product manufacturer but also to attract the attention of potential buyers. Manufacturers carefully design their logos so that their characteristics, impressions and philosophies are expressed. Moreover, logos on a person’s belongings can play an important role in characterizing and identifying the person. Extraction and recognition of logos from images captured by multiple surveillance cameras could provide useful information for identification and tracking of individuals. Automatic extraction and recognition of shoe logos using image analysis techniques is challenging because they have characteristics that distinguish them from those of other products, and their appearance can vary substantially. Automatic extraction and recognition techniques must handle these problems properly, because shoes are usually worn on feet and,therefore, move frequently with respect to stationary cameras. Additionally, since shoe logos are usually appeared as an integrated design component of shoe design, they have significant within-class appearance variation due to the variation in color, fabric material, shape of shoe and dirt or aging of shoe. The extraction and recognition of logos in images has attracted the attention of many researchers. Several studies on the automatic extraction of logos have been reported. Affine and non-rigid transformation occurs frequently in real-world images. This makes logo detection and recognition complex, especially for model-based approaches, owing to the difficulty in collecting sufficient samples to obtain a robust model. Several related works [8, 9, 14, 15] address this problem. While these related works were carefully designed for handling appearance variations, they did not pay enough attention for large within-class appearance variation. Therefore, extending the target of these methods for shoe logos is quite difficult. Moreover,deep neural network architectures have been employed in image recognition tasks due to their promising performance and high adaptivity. However a deep learning method requires a large-scale, accurately annotated dataset for training. Creating such a dataset for deep learning is difficult because shoe logos have a wide variety of appearances even in the same bland. In the present paper, we propose an automatic extraction and recognition method for shoe logos using a limited number of training samples. The proposed method employs maximally stable extremal regions (MSERs) [4] for the initial region extraction, an iterative algorithm for region grouping and gradient features, and two-stage support vector machines (SVM) for logo recognition. For performance evaluation, we use the IEICE-PRMU shoe logo dataset which consists of shoe logo images captured in uncontrolled condition. The results of performance evaluation experiments show that the proposed method achieves promising performance for both logo extraction and recognition. In the present paper, Chapter 1 gives the introduction of this research. Chapter 2 expresses related works. Chapter 3 shows the proposed method in detail. Chapter 4 represents evaluation experiments. In the last Chapter, a conclusion is explained. Appendix A submits the method which proposed methods are employed. Appendix B describes individual diversification extraction strategies. Postscripts shows where the program is located and presentation side which is used on master research presentation.
内容記述
内容記述タイプ Other
内容記述 Human Interface Laboratory Division of Information Engineering Graduate School of Engineering Mie University
内容記述
内容記述タイプ Other
内容記述 48p
書誌情報
発行日 2018-03
フォーマット
内容記述タイプ Other
内容記述 application/pdf
著者版フラグ
出版タイプ VoR
出版タイプResource http://purl.org/coar/version/c_970fb48d4fbd8a85
出版者
出版者 三重大学
出版者(ヨミ)
値 ミエダイガク
修士論文指導教員
寄与者識別子Scheme WEKO
寄与者識別子 40077
姓名 若林, 哲史
言語 ja
資源タイプ(三重大)
値 Master's Thesis / 修士論文
戻る
0
views
See details
Views

Versions

Ver.1 2023-06-19 15:54:01.209803
Show All versions

Share

Mendeley Twitter Facebook Print Addthis

Cite as

エクスポート

OAI-PMH
  • OAI-PMH JPCOAR 2.0
  • OAI-PMH JPCOAR 1.0
  • OAI-PMH DublinCore
  • OAI-PMH DDI
Other Formats
  • JSON
  • BIBTEX

Confirm


Powered by WEKO3


Powered by WEKO3