Presentation 2006/1/3
APPLICATION OF STATISTICAL MODELS TO ESTIMATE PERSONAL TASTES IN TEXTILE DESIGN(International Workshop on Advanced Image Technology 2006)
Kotaro Nakatani, Kosuke Moriwaki, Yoshihiro Kamei,
PDF Download Page PDF download Page Link
Abstract(in Japanese) (See Japanese page)
Abstract(in English) Textile design companies should be aware of consumer tastes regarding brand-new designs. Inquiry is the only method available to gather this information. This paper describes a new technique for estimating personal tastes regarding textile design. For machine learning, we apply two statistical models-nonlinear Support Vector Machine (SVM) and linear discriminant function-that learn the relationship between each design's image features and personal tastes and classify the design samples as like or dislike. Processing each textile image yields twelve color and three shape features influencing impressions. We estimate recognition accuracy based on the data comprising twenty-one samples and eleven personal tastes. We select one sample and construct models using the others. The selected sample is classified using the models, and their recognition accuracies are compared. The result indicates that the nonlinear SVM can estimate personal tastes more accurately than the linear discriminant function; their mean recognition rates are 86.4% and 64.1%, respectively.
Keyword(in Japanese) (See Japanese page)
Keyword(in English)
Paper # IE2005-214
Date of Issue

Conference Information
Committee IE
Conference Date 2006/1/3(1days)
Place (in Japanese) (See Japanese page)
Place (in English)
Topics (in Japanese) (See Japanese page)
Topics (in English)
Chair
Vice Chair
Secretary
Assistant

Paper Information
Registration To Image Engineering (IE)
Language ENG
Title (in Japanese) (See Japanese page)
Sub Title (in Japanese) (See Japanese page)
Title (in English) APPLICATION OF STATISTICAL MODELS TO ESTIMATE PERSONAL TASTES IN TEXTILE DESIGN(International Workshop on Advanced Image Technology 2006)
Sub Title (in English)
Keyword(1)
1st Author's Name Kotaro Nakatani
1st Author's Affiliation Department of Information and Electronics Technology Research Institute of Osaka Prefecture()
2nd Author's Name Kosuke Moriwaki
2nd Author's Affiliation Department of Information and Electronics Technology Research Institute of Osaka Prefecture
3rd Author's Name Yoshihiro Kamei
3rd Author's Affiliation Cozy Design
Date 2006/1/3
Paper # IE2005-214
Volume (vol) vol.105
Number (no) 501
Page pp.pp.-
#Pages 5
Date of Issue