FoCo system: a tool to bridge the domain gap between fashion and artificial intelligence. Issue 5 (2nd September 2019)

• Record Type: Journal Article
• Authors: Zou, Xingxing; Wong, Wai Keung; Gao, Can; Zhou, Jie
• Is Part Of: International journal of clothing science and technology. Volume 31:Issue 5(2019)
• Page Start: 630
• Page End: 643
• Publication Date: 2019-09-02
• Subjects: Fashion -- Artificial intelligence -- Color cognition -- Color matching -- Style analysis; Clothing and dress -- Periodicals; Textile fabrics -- Periodicals; 677
• DOI: 10.1108/IJCST-10-2018-0130
• Languages: English
• ISSNs: 0955-6222

Abstract

Abstract : Purpose: The deficiency of the mapping between fashion color (FoCo) value and linguistic color expression causes the difficulty of machine-based fashion understanding tasks that are heavily associated with color matching. The purpose of this paper is to propose the FoCo system and construct it with four steps, in order to bridge this gap. Design/methodology/approach: The color distribution in HSB color space is analyzed to estimate the rough number of color categories. Similar color values are grouped to obtain the initial HSB value range for each color category. The intra-category color differences are calculated to determine their final HSB value ranges and Pantone color is used for fine-tuning. Findings: With practical applications in mind, the FoCo system is designed as a hierarchical structure with three layers. Originality/value: The FoCo system is designed as a hierarchical structure with three layers: color units for color matching-related tasks, color categories for style analysis tasks and color tones for color recognition tasks. Extensive experiments demonstrate the effectiveness of the FoCo system.