Danbooru2020

Danbooru2020是一个大型动画图像数据库，包含420万多个图像，并附有1.3亿多个详细描述图像内容的文本注释。该数据集可用于机器学习，例如图像识别和生成。该数据集已用于各种应用程序，尤其是生成建模。
Danbooru20xx每年都会根据前几年的图像和元数据进行改进和更新。以前的版本是: Danbooru2017、Danbooru2018、danbooru2019。

Danbooru2020 is a large-scale animated image database containing over 4.2 million images paired with more than 130 million detailed text annotations that describe the content of the images. This dataset can be used for machine learning tasks such as image recognition and generation, and has been applied to various applications, especially generative modeling.
Danbooru20xx is improved and updated annually based on the images and metadata from previous years. Its prior versions include Danbooru2017, Danbooru2018 and Danbooru2019.

搜集汇总

背景与挑战

背景概述

Danbooru2020是一个大型动画图像数据库，包含420万张图像和1.3亿条文本注释，适用于图像识别与生成等机器学习任务。该数据集是Danbooru系列（如2017-2019版本）的年度更新版本，于2021年发布。