LonelVino/EditBench

EditBench是一个全面的图像编辑数据集，旨在促进图像编辑领域的研究和开发。它提供了多样化的图像、编辑类型和注释，以系统评估图像编辑模型。数据集包含约40,000张图像，分辨率为1024x1024、768x1360和1360x768，涵盖17种不同的编辑类型。每张图像都有原始提示、编辑指令、编辑后的提示、编辑主题、掩码图像和评估分数等注释。数据集结构包括图像、掩码和元数据JSON文件，分别存储原始和编辑后的图像、对应的掩码图像以及每个编辑类型的元数据。

EditBench is a comprehensive image editing dataset designed to facilitate research and development in the field of image editing. It contains about 40,000 images with resolutions of 1024x1024, 768x1360, and 1360x768, covering 17 distinct editing types. The dataset provides annotations such as original prompts, editing instructions, edited prompts, edit subjects, mask images, and evaluation scores. The dataset structure includes image folders, mask folders, and metadata JSON files.