Bin1117/AnyEdit

AnyEdit是一个全面的多模态指令编辑数据集，包含250万高质量的编辑对，涵盖25种编辑类型和五个领域。该数据集主要用于基于指令的图像编辑研究，支持文本到图像和图像到图像的转换。数据集的结构包括训练集、验证集和测试集，每个样本包含图像ID、编辑指令、编辑类型、原始图像文件、编辑后的图像文件、视觉输入、输入文本和输出文本等信息。

AnyEdit is a comprehensive multi-modal instruction editing dataset, comprising 2.5 million high-quality editing pairs spanning 25 editing types and five domains. The dataset categorizes image editing tasks into five groups based on different editing capabilities: local editing, global editing, camera move editing, implicit editing, and visual editing. The dataset structure includes training and validation sets, along with detailed instruction formats and file structures. The primary use of the dataset is for research on text-to-image and instruction-based image editing.