EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Revealing Gender Bias from Prompt to Image in Stable Diffusion

Yankun Wu, Yuta Nakashima, Noa Garcia

Journal of Imaging · 24.01.2025 · 11 (2) · 35

DOI: 10.3390/jimaging11020035 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Journal of Imaging, online erschienen am 24.01.2025. Verfasst von Yankun Wu und 2 weiteren Autor:innen. Beteiligte Einrichtung: D3 Center, Osaka University, Suita 565-0871, Osaka, Japan. Die Arbeit wurde bisher 13-mal zitiert (Crossref) und nennt 83 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: JST CREST, JST FOREST, JSPS KAKENHI.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Social biases in generative models have gained increasing attention. This paper proposes an automatic evaluation protocol for text-to-image generation, examining how gender bias originates and perpetuates in the generation process of Stable Diffusion. Using triplet prompts that vary by gender indicators, we trace presentations at several stages of the generation process and explore dependencies between prompts and images. Our findings reveal the bias persists throughout all internal stages of the generating process and manifests in the entire images. For instance, differences in object presence, such as different instruments and outfit preferences, are observed across genders and extend to overall image layouts. Moreover, our experiments demonstrate that neutral prompts tend to produce images more closely aligned with those from masculine prompts than with their female counterparts. We also investigate prompt-image dependencies to further understand how bias is embedded in the generated content. Finally, we offer recommendations for developers and users to mitigate this effect in text-to-image generation.

Schlagwörter

Stichwörter der Autor:innen

bias evaluationgender biastext-to-image generation

Autor:innen und Institutionen

  • D3 Center, Osaka University, Suita 565-0871, Osaka, Japan.

Förderung

  • JST CREST
  • JST FOREST
  • JSPS KAKENHI
  • JST SPRING

Zitieren

APA

Wu, Y., Nakashima, Y., & Garcia, N. (2025). Revealing Gender Bias from Prompt to Image in Stable Diffusion. Journal of Imaging, 11(2), 35. https://doi.org/10.3390/jimaging11020035

Vancouver

Wu Y, Nakashima Y, Garcia N. Revealing Gender Bias from Prompt to Image in Stable Diffusion. Journal of Imaging. 2025;11(2):35. doi:10.3390/jimaging11020035

BibTeX
@article{Wu2025,
  author = {Wu, Yankun and Nakashima, Yuta and Garcia, Noa},
  title = {Revealing Gender Bias from Prompt to Image in Stable Diffusion},
  journal = {Journal of Imaging},
  year = {2025},
  volume = {11},
  number = {2},
  pages = {35},
  doi = {10.3390/jimaging11020035}
}
Kostenlos starten →