SGDiff: A Style Guided Diffusion Model for Fashion Synthesis

Zhengwentai Sun, Yanghong Zhou, Honghong He, P. Y. Mok *
The Hong Kong Polytechnic University
ACM Multimedia 2023
*Indicates Corresponding Author
Paper Supplementary Code arXiv

Illustration of the challenges in synthesizing images of fashion clothing with specific attributes and artistic styles: the SGDiff results of (d) are compared with the results based on text-only descriptions of (b) and (c).

Abstract

This paper reports on the development of a novel style guided diffusion model (SGDiff) which overcomes certain weaknesses inherent in existing models for image synthesis. The proposed SGDiff combines image modality with a pretrained text-to-image diffusion model to facilitate creative fashion image synthesis. It addresses the limitations of text-to-image diffusion models by incorporating supplementary style guidance, substantially reducing training costs, and overcoming the difficulties of controlling synthesized styles with text-only inputs. This paper also introduces a new dataset -- SG-Fashion, specifically designed for fashion image synthesis applications, offering high-resolution images and an extensive range of garment categories. By means of comprehensive ablation study, we examine the application of classifier-free guidance to a variety of conditions and validate the effectiveness of the proposed model for generating fashion images of the desired categories, product attributes, and styles. The contributions of this paper include a novel classifier-free guidance method for multi-modal feature fusion, a comprehensive dataset for fashion image synthesis application, a thorough investigation on conditioned text-to-image synthesis, and valuable insights for future research in the text-to-image synthesis domain. The code and dataset will be released once the paper is published.

Methods

The proposed Style-Guided Diffusion Model (SGDiff) network (a) an overview and (b) detail model: SGDiff takes two inputs, a text condition (ct) for garment attributes and a style condition (cs) for style guidance, and leverages the Skip Cross-Attention (SCA) module and a pretrained CLIP image encoder for efficient training and resource utilization.

BibTeX


      @inproceedings{10.1145/3581783.3613806,
        author = {Sun, Zhengwentai and Zhou, Yanghong and He, Honghong and Mok, P. Y.},
        title = {SGDiff: A Style-Guided Diffusion Model for Fashion Synthesis},
        year = {2023},
        isbn = {979-8-4007-0108-5/23/10},
        publisher = {Association for Computing Machinery},
        address = {New York, NY, USA},
        url = {https://doi.org/10.1145/3581783.3613806},
        doi = {10.1145/3581783.3613806},
        booktitle = {Proceedings of the 31th ACM International Conference on Multimedia},
        numpages = {10},
        keywords = {fashion synthesis, style guidance, text-to-image, denoising diffusion probabilistic models},
        location = {Ottawa, ON, Canada},
        series = {MM '23}
        }