CyberHarem/skyfire_arknights

这是一个名为skyfire/スカイフレア/天火 (Arknights)的数据集，包含356张图片及其标签。图片从多个网站（如danbooru、pixiv、zerochan等）爬取，爬取系统由DeepGHS团队开发。数据集的核心标签包括长头发、动物耳朵、棕色头发、猫耳朵等。文件中还列出了数据集的多个包及其下载链接，并提供了使用waifuc加载原始数据集的代码示例。此外，文件还展示了标签聚类结果的列表。This is a dataset named skyfire/スカイフレア/天火 (Arknights), which contains 356 images and their associated tags. The images were crawled from various platforms including danbooru, pixiv, zerochan, and others, and the crawling system was developed by the DeepGHS team. The core tags of this dataset include long hair, animal ears, brown hair, cat ears, and more. The document also lists multiple packages of the dataset along with their download links, and provides a code example for loading the raw dataset using waifuc. Furthermore, it presents a list of tag clustering results.

原始信息汇总

数据集概述

数据集基本信息

• 名称: Dataset of skyfire/スカイフレア/天火 (Arknights)
• 许可证: MIT
• 任务类别: text-to-image
• 标签: art, not-for-all-audiences
• 大小类别: n<1K

数据集内容

• 图像数量: 356张
• 核心标签: long_hair, animal_ears, brown_hair, cat_ears, breasts, brown_eyes, tail, cat_tail, cat_girl, earrings, medium_breasts, large_breasts, hair_between_eyes

数据集包列表

名称	图像数量	大小	类型	描述
raw	356	578.27 MiB	Waifuc-Raw	原始数据，包含元信息（最小边对齐到1400像素，如果更大）
1200	356	487.72 MiB	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	912	918.16 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	15		1girl, black_pantyhose, black_skirt, long_sleeves, open_jacket, pleated_skirt, solo, belt, red_jacket, white_shirt, collared_shirt, looking_at_viewer, wide_sleeves, closed_mouth, white_background, black_necktie, holding_staff, simple_background, jewelry, fire, smile, cowboy_shot
1	5		1girl, belt, black_footwear, black_necktie, black_skirt, full_body, gradient_legwear, high_heels, holding_staff, long_sleeves, looking_at_viewer, miniskirt, open_jacket, pleated_skirt, red_jacket, wide_sleeves, black_pantyhose, closed_mouth, simple_background, smile, solo, white_shirt, black_background, brown_pantyhose, collared_shirt, standing, black_shirt, buckle, fire, orange_eyes, shoes
2	9		1girl, belt, black_footwear, black_skirt, brown_pantyhose, collared_shirt, gradient_legwear, long_sleeves, looking_at_viewer, open_jacket, red_jacket, simple_background, solo, white_background, white_shirt, black_necktie, full_body, high_heels, wide_sleeves, black_pantyhose, closed_mouth, dress, shoes, smile, pleated_skirt, blush, holding_staff, jewelry, orange_eyes, red_pantyhose, hand_up, standing
3	10		1girl, black_necktie, collared_shirt, long_sleeves, red_jacket, solo, upper_body, white_shirt, jewelry, looking_at_viewer, open_jacket, blush, closed_mouth, simple_background, smile, white_background, black_jacket, black_vest, orange_eyes, piercing, wide_sleeves
4	7		1girl, bare_shoulders, black_one-piece_swimsuit, blush, looking_at_viewer, simple_background, solo, white_background, ear_piercing, jewelry, official_alternate_costume, closed_mouth, very_long_hair, arm_strap, casual_one-piece_swimsuit, o-ring_swimsuit, sideboob

搜集汇总

背景与挑战

背景概述

该数据集名为'CyberHarem/skyfire_arknights'，由CyberHarem组织创建，专注于文本到图像生成任务，规模在1K到10K之间，属于艺术类别。数据集标记为'Not-For-All-Audiences'，包含敏感内容，使用MIT许可证，适合研究敏感艺术内容生成，但用户需谨慎处理潜在风险。