The article presents a new dataset, FIRST, designed for text-driven fashion synthesis and design. It’s part of the artificial intelligence generative content (AIGC) that can bring a significant revolution in the traditional fashion industry. The dataset includes a million high-resolution fashion images with detailed textual descriptions. It covers a wide range of attire categories, and the image-paired textual descriptions are organized at multiple hierarchical levels. The authors highlight the importance of this dataset for developing more intelligent fashion synthesis and design systems that can enhance creativity and imagination in fashion design.
Publication date: 13 Nov 2023
Project Page: https://arxiv.org/abs/2311.07414v1
Paper: https://arxiv.org/pdf/2311.07414
