Construction of a Dataset Including Actual Sewing Patterns for Clothing Design and Sewing Pattern Generation Using GAN

In the process of garment production, creating sewing patterns which serve as design blueprints for clothing, from fashion sketches is a highly complex task that requires extensive experience and knowledge from garment makers. In this paper, we propose a method for generating sewing patterns from clothing images by constructing a dataset based on sewing patterns used in actual dressmaking.In addition, we train a pattern image generation model using Generative Adversarial Networks and report that In some cases, it was possible to generate sewing patterns that reflected the shape of the clothing.