CoCoStyle: Mixed initiative co-creative system to support creative process of fashion design

Abstract

Due to the rapid development of AI, research on creativity support tools that allow humans to foster creativity through collaboration with AI is actively progressing. The concept of an AI-based mixed initiative co-creative system (AI-MICCS), which gives the user the initiative, has emerged. Our research focuses on the fashion domain, where creativity plays an important role, and we present CoCoStyle, an AI-MICCS that can support the creativity of fashion designers. CoCoStyle has three main phases (CoResearch, CoDesign, and CoImprovement) and six sub-phases (Build, Pick, Create, Switch, View and Fit). Fashion Designers can efficiently group a large number of desired images using the attribute detection model and k-means algorithm, create unique designs using StyleGAN2, and improve designs from various perspectives using transformer embedding and attribute detection results. We believe that CoCoStyle will effectively support fashion designers in their creative design process by allowing users to take the initiative in controlling the AI.