jobs-git/diffusiondb

DiffusionDB是一个大规模的文本到图像提示数据集，包含1400万由Stable Diffusion模型根据真实用户指定的提示和超参数生成的图像。该数据集旨在帮助研究人员理解提示和生成模型之间的相互作用，检测深度伪造，并设计人机交互工具，以便用户更容易地使用这些模型。数据集提供了两个子集：DiffusionDB 2M和DiffusionDB Large，以满足不同的研究需求。数据集的文件结构被模块化，以便于分发和加载。每个图像都有一个JSON文件链接到其提示和超参数。

DiffusionDB is a large-scale text-to-image prompt dataset containing 14 million images generated by Stable Diffusion using prompts and hyperparameters specified by real users. It aims to assist researchers in understanding the interplay between prompts and generative models, detecting deepfakes, and designing human-AI interaction tools. The dataset is available in two subsets: DiffusionDB 2M and DiffusionDB Large, catering to different research needs. The file structure is modularized for easy distribution and loading, with each image linked to its prompts and hyperparameters via a JSON file.