A Deep Learning-Based Approach to Facilitate Cross-cultural Kansei Design

Xiaofei Zhou, Pei Luen Patrick Rau, Michiko Ohkura, Tipporn Laohakangvalvit, Bingcheng Wang

研究成果: Conference contribution

抄録

With the development of economic globalization, more and more product designers are faced with the need of designing for customers from different countries. However, it is a challenge for designers to efficiently develop the desired mental images of certain products for target users with different cultural backgrounds. Our research proposed a deep learning-based system to facilitate designers to gain better awareness of the cross-culture differences between different target customers. We trained a kawaii classification neural network model with the data of 1414 cosmetic packaging images annotated by 12 Japanese females separately. As a follow-up investigation, we conducted neuron analysis to compare the features of kawaii packages perceived by Japanese participants with the results from a prior study conducted with Chinese participants. The result shows that Japanese females tended to see more girlish and exquisite design features as kawaii while Chinese females perceived more childish and round elements as kawaii. A reverse experiment further verified the effectiveness of adding these different design features to enhance Chinese or Japanese females’ perception of kawaii. We also noticed that it’s hard to obtain the cross-cultural differences in customers’ perception by extracting image parameters with a set of predefined visual features as such perception differences could be subconscious. Our deep learning-based Kansei design facilitation provides a feasible solution to customized design for target customers with different cultural backgrounds.

本文言語English
ホスト出版物のタイトルCross-Cultural Design. Interaction Design Across Cultures - 14th International Conference, CCD 2022, Held as Part of the 24th HCI International Conference, HCII 2022, Proceedings
編集者Pei-Luen Patrick Rau
出版社Springer Science and Business Media Deutschland GmbH
ページ145-160
ページ数16
ISBN(印刷版)9783031060373
DOI
出版ステータスPublished - 2022
イベント14th International Conference on Cross-Cultural Design, CCD 2022 Held as Part of the 24th HCI International Conference, HCII 2022 - Virtual, Online
継続期間: 2022 6月 262022 7月 1

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)
13311 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Conference

Conference14th International Conference on Cross-Cultural Design, CCD 2022 Held as Part of the 24th HCI International Conference, HCII 2022
CityVirtual, Online
Period22/6/2622/7/1

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

「A Deep Learning-Based Approach to Facilitate Cross-cultural Kansei Design」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