PD12M 大规模图像-文本数据集
收藏资源简介:
Public Domain 12M(简称 PD12M)是由 Spawning 于 2024 年创建的一个大规模图像-文本数据集,它包含了 1240 万张高质量的公共领域及 CC0 许可图片,这些图片搭配了合成字幕,主要用于训练文本到图像的模型。 PD12M 是目前最大的公共领域图像-文本数据集,以其庞大的规模和明确的版权声明,为 AI 模型的训练提供了坚实的基础,同时最小化了版权担忧。相关论文成果为「Public Domain 12M: A Highly Aesthetic Image-Text Dataset with Novel Governance Mechanisms」。
Public Domain 12M (abbreviated as PD12M) is a large-scale image-text dataset developed by Spawning in 2024. It comprises 12.4 million high-quality public domain and CC0-licensed images paired with synthetic captions, and is primarily intended for training text-to-image models. PD12M currently stands as the largest public domain image-text dataset, offering a robust foundation for AI model training through its massive scale and explicit copyright statements, while effectively mitigating copyright-related concerns. The associated research paper is titled "Public Domain 12M: A Highly Aesthetic Image-Text Dataset with Novel Governance Mechanisms"




