Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Towards Intelligent Design: A Self-driven Framework for Collocated Clothing Synthesis Leveraging Fashion Styles and Textures

Minglong Dong, Dongliang Zhou, Jianghong Ma, Haijun Zhang

TL;DR

This work tackles collocated clothing synthesis (CCS) without paired outfits by introducing ST-Net, a self-supervised GAN-based framework that learns fashion compatibility from unpaired clothing images using style and texture cues. Built on GAN inversion with pre-trained StyleGANs for upper and lower garments, ST-Net incorporates a style- and texture-guided discriminator ${D}_{ST}$ and a dual discriminator ${D}_{dual}$ to enhance realism and stabilize training. The model is trained with a composite loss that aligns style via color histograms and texture via instance-level patch discrimination, while enforcing compatibility between input and generated items. A large-scale unlabeled CCS dataset supports evaluation, and experiments show ST-Net outperforms state-of-the-art unsupervised I2I baselines in both visual authenticity (FID) and fashion compatibility (FCTS), indicating strong practical potential for automatic outfit synthesis.

Abstract

Collocated clothing synthesis (CCS) has emerged as a pivotal topic in fashion technology, primarily concerned with the generation of a clothing item that harmoniously matches a given item. However, previous investigations have relied on using paired outfits, such as a pair of matching upper and lower clothing, to train a generative model for achieving this task. This reliance on the expertise of fashion professionals in the construction of such paired outfits has engendered a laborious and time-intensive process. In this paper, we introduce a new self-driven framework, named style- and texture-guided generative network (ST-Net), to synthesize collocated clothing without the necessity for paired outfits, leveraging self-supervised learning. ST-Net is designed to extrapolate fashion compatibility rules from the style and texture attributes of clothing, using a generative adversarial network. To facilitate the training and evaluation of our model, we have constructed a large-scale dataset specifically tailored for unsupervised CCS. Extensive experiments substantiate that our proposed method outperforms the state-of-the-art baselines in terms of both visual authenticity and fashion compatibility.

Towards Intelligent Design: A Self-driven Framework for Collocated Clothing Synthesis Leveraging Fashion Styles and Textures

TL;DR

This work tackles collocated clothing synthesis (CCS) without paired outfits by introducing ST-Net, a self-supervised GAN-based framework that learns fashion compatibility from unpaired clothing images using style and texture cues. Built on GAN inversion with pre-trained StyleGANs for upper and lower garments, ST-Net incorporates a style- and texture-guided discriminator and a dual discriminator to enhance realism and stabilize training. The model is trained with a composite loss that aligns style via color histograms and texture via instance-level patch discrimination, while enforcing compatibility between input and generated items. A large-scale unlabeled CCS dataset supports evaluation, and experiments show ST-Net outperforms state-of-the-art unsupervised I2I baselines in both visual authenticity (FID) and fashion compatibility (FCTS), indicating strong practical potential for automatic outfit synthesis.

Abstract

Collocated clothing synthesis (CCS) has emerged as a pivotal topic in fashion technology, primarily concerned with the generation of a clothing item that harmoniously matches a given item. However, previous investigations have relied on using paired outfits, such as a pair of matching upper and lower clothing, to train a generative model for achieving this task. This reliance on the expertise of fashion professionals in the construction of such paired outfits has engendered a laborious and time-intensive process. In this paper, we introduce a new self-driven framework, named style- and texture-guided generative network (ST-Net), to synthesize collocated clothing without the necessity for paired outfits, leveraging self-supervised learning. ST-Net is designed to extrapolate fashion compatibility rules from the style and texture attributes of clothing, using a generative adversarial network. To facilitate the training and evaluation of our model, we have constructed a large-scale dataset specifically tailored for unsupervised CCS. Extensive experiments substantiate that our proposed method outperforms the state-of-the-art baselines in terms of both visual authenticity and fashion compatibility.
Paper Structure (9 sections, 6 equations, 2 figures, 1 table)

This paper contains 9 sections, 6 equations, 2 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methodology
  3. Proposed Framework
  4. Training Losses
  5. Experiments
  6. Implemention Details
  7. Comparison to State-of-the-art Methods
  8. Ablation Study
  9. Conclusion

Figures (2)

  • Figure 1: Overview of our proposed ST-Net: (a) the training phase of $D_{ST}$ and (b) the training phase of the generator $G$.
  • Figure 2: Comparison between ST-Net and other baselines, which are CycleGAN zhu2017unpaired, MUNIT huang2018munit, DRIT DRIT, DRIT++ DRIT_plus, StarGAN-v2 choi2020stargan, and GP-UNIT yang2023gp: (a) 'upper $\rightarrow$ lower' direction and (b) 'lower $\rightarrow$ upper' direction.