CyberHarem/mifune_miyu_idolmastercinderellagirls

该数据集是关于《偶像大师：灰姑娘女孩》中的角色三船美優（Mifune Miyu）的图像数据集，包含487张图像及其标签。图像从多个网站（如danbooru、pixiv、zerochan等）爬取，爬取系统由DeepGHS团队提供支持。数据集的核心标签包括`brown_hair, long_hair, breasts, brown_eyes, ponytail, bangs, large_breasts`，这些标签在数据集中被修剪。数据集提供了多种不同尺寸和裁剪方式的图像包，如raw、800、stage3-p480-800、1200和stage3-p480-1200。此外，还提供了使用waifuc加载原始数据集的代码示例，以及标签聚类结果的列表。

This is an image dataset centered on the character Mifune Miyu from *The Idolmaster: Cinderella Girls*, containing 487 images and their corresponding tags. The images were crawled from multiple platforms including danbooru, pixiv, zerochan and others, with the crawling system powered by the DeepGHS team. The core tags of the dataset are `brown_hair, long_hair, breasts, brown_eyes, ponytail, bangs, large_breasts`, and these tags are pruned within the dataset. Multiple image packages with different sizes and cropping methods are provided, such as raw, 800, stage3-p480-800, 1200 and stage3-p480-1200. Additionally, code examples for loading the raw dataset using waifuc, as well as a list of tag clustering results, are provided.

原始信息汇总

数据集概述

数据集基本信息

• 名称: mifune_miyu/三船美優 (THE iDOLM@STER: Cinderella Girls)
• 许可证: MIT
• 任务类别: text-to-image
• 标签: art, not-for-all-audiences
• 规模: n<1K

数据集内容

• 图像数量: 487张
• 核心标签: brown_hair, long_hair, breasts, brown_eyes, ponytail, bangs, large_breasts
• 图像来源: 从多个网站爬取，如danbooru, pixiv, zerochan等

数据集包列表

名称	图像数量	大小	类型	描述
raw	487	522.91 MiB	Waifuc-Raw	包含元信息的原始数据（最小边对齐到1400像素，如果更大）
800	487	342.76 MiB	IMG+TXT	短边不超过800像素的数据集
stage3-p480-800	1117	684.74 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素
1200	487	476.56 MiB	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	1117	899.28 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素

标签聚类结果

原始文本版本

#	样本数量	图像示例	标签
0	7		1girl, blush, cleavage, looking_at_viewer, smile, solo, bikini, navel, collarbone, jewelry
1	6		1girl, looking_at_viewer, necklace, smile, solo, blush, simple_background, white_background, bare_shoulders, collarbone, bracelet, sleeveless_dress
...	...	...	...

表格版本

#	样本数量	图像示例	1girl	blush	cleavage	looking_at_viewer	smile	solo	...
0	7		X	X	X	X	X	X	...
1	6		X	X		X	X	X	...
...	...	...	...	...	...	...	...	...	...

搜集汇总

背景与挑战

背景概述

该数据集名为'mifune_miyu_idolmastercinderellagirls'，专注于文本到图像生成任务，规模中等（1K-10K），包含艺术相关且可能敏感的内容，采用MIT许可证。