Automatic attribute discovery with neural activations

Sirion Vittayakorn, Takayuki Umeda, Kazuhiko Murasaki, Kyoko Sudo, Takayuki Okatani, Kota Yamaguchi

研究成果: Conference contribution

22 被引用数 (Scopus)

抄録

How can a machine learn to recognize visual attributes emerging out of online community without a definitive supervised dataset? This paper proposes an automatic approach to discover and analyze visual attributes from a noisy collection of image-text data on the Web. Our approach is based on the relationship between attributes and neural activations in the deep network. We characterize the visual property of the attribute word as a divergence within weakly-annotated set of images. We show that the neural activations are useful for discovering and learning a classifier that well agrees with human perception from the noisy real-world Web data. The empirical study suggests the layered structure of the deep neural networks also gives us insights into the perceptual depth of the given word. Finally, we demonstrate that we can utilize highly-activating neurons for finding semantically relevant regions.

本文言語English
ホスト出版物のタイトルComputer Vision - 14th European Conference, ECCV 2016, Proceedings
編集者Bastian Leibe, Jiri Matas, Nicu Sebe, Max Welling
出版社Springer Verlag
ページ252-268
ページ数17
ISBN(印刷版)9783319464923
DOI
出版ステータスPublished - 2016
イベント14th European Conference on Computer Vision, ECCV 2016 - Amsterdam, Netherlands
継続期間: 2016 10月 82016 10月 16

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
9908 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Conference

Conference14th European Conference on Computer Vision, ECCV 2016
国/地域Netherlands
CityAmsterdam
Period16/10/816/10/16

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)

フィンガープリント

「Automatic attribute discovery with neural activations」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル