ORES: Open-vocabulary Responsible Visual Synthesis

Open-vocabulary Responsible Visual Synthesis (ORES) aims to generate an image under the user's query u that meets two criteria: 1) it should not contain a specified visual concept, represented as c, which is defined by the administrator in practice, and 2) it should undergo minimal changes compared to the image Iu generated by original user query directly.

Data and Resources

Cite this as

Minheng Ni, Chenfei Wu, Xiaodong Wang, Shengming Yin, Lijuan Wang, Zicheng Liu, Nan Duan (2024). Dataset: ORES: Open-vocabulary Responsible Visual Synthesis. https://doi.org/10.57702/f8fbh4un

DOI retrieved: December 16, 2024

Additional Info

Field Value
Created December 16, 2024
Last update December 16, 2024
Defined In https://doi.org/10.48550/arXiv.2308.13785
Author Minheng Ni
More Authors
Chenfei Wu
Xiaodong Wang
Shengming Yin
Lijuan Wang
Zicheng Liu
Nan Duan
Homepage https://github.com/kodenii/ORES