CyberHarem/hinata_bluearchive

该数据集是关于Blue Archive中的角色hinata/若葉ヒナタ/日向的图像数据集，包含476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长头发、黑头发、胸部、光环、头发遮住一只眼、大胸部、红眼睛、耳环、十字耳环、非常长的头发、帽子、太阳帽等。数据集提供了多种格式的下载，包括原始数据、1200像素限制的数据集以及三阶段裁剪的数据集。此外，还提供了使用waifuc加载原始数据集的代码示例。

This is an image dataset featuring the character Hinata (Wakaba Hinata / 日向) from Blue Archive, containing 476 images and their corresponding labels. The images were crawled from multiple platforms such as Danbooru, Pixiv, Zerochan and other sites, with the crawling system powered by the DeepGHS Team. Core tags of the dataset include long hair, black hair, chest, halo, hair over one eye, large breasts, red eyes, earrings, cross earrings, very long hair, hat, sun hat and others. Various download formats are provided, including the raw dataset, the 1200-pixel restricted dataset, and the three-stage cropped dataset. Additionally, a code example for loading the raw dataset using waifuc is also available.

原始信息汇总

数据集概述：hinata/若葉ヒナタ/日向 (Blue Archive)

数据集基本信息

• 名称: hinata/若葉ヒナタ/日向 (Blue Archive)
• 包含内容: 476张图像及其标签
• 核心标签: long_hair, black_hair, breasts, halo, hair_over_one_eye, large_breasts, red_eyes, earrings, cross_earrings, very_long_hair, hat, sun_hat
• 许可证: MIT
• 任务类别: text-to-image
• 标签: art, not-for-all-audiences
• 大小类别: n<1K

数据集包信息

名称	图像数量	大小	类型	描述
raw	476	804.41 MiB	Waifuc-Raw	包含元信息的原始数据（如果较大，则最小边缘对齐到1400）
1200	476	671.65 MiB	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	1259	1.36 GiB	IMG+TXT	三阶段裁剪数据集，区域不小于480x480像素

数据集加载示例

python import os import zipfile

from huggingface_hub import hf_hub_download from waifuc.source import LocalSource

下载原始存档文件

zip_file = hf_hub_download( repo_id=CyberHarem/hinata_bluearchive, repo_type=dataset, filename=dataset-raw.zip, )

提取文件到你的目录

dataset_dir = dataset_dir os.makedirs(dataset_dir, exist_ok=True) with zipfile.ZipFile(zip_file, r) as zf: zf.extractall(dataset_dir)

使用waifuc加载数据集

source = LocalSource(dataset_dir) for item in source: print(item.image, item.meta[filename], item.meta[tags])

数据集标签聚类结果

聚类列表

#	样本数	标签
0	8	1girl, asymmetrical_bangs, blush, cleavage_cutout, cross, habit, jewelry, nun, simple_background, solo, white_background, eyes_visible_through_hair, looking_at_viewer, long_sleeves, upper_body, closed_mouth, necktie
1	6	1girl, asymmetrical_bangs, cleavage_cutout, cross, habit, jewelry, long_sleeves, looking_at_viewer, nun, pelvic_curtain, simple_background, solo, white_background, white_thighhighs, blush, closed_mouth, sitting, smile, necktie
2	8	1girl, black_choker, casual_one-piece_swimsuit, cleavage, closed_mouth, collarbone, looking_at_viewer, official_alternate_costume, solo, white_one-piece_swimsuit, blush, covered_navel, smile, bare_shoulders, cowboy_shot, hat_flower, braid, one_eye_covered, yellow_headwear, criss-cross_halter, jewelry, simple_background, white_background, beach, yellow_halo
3	7	1girl, bare_shoulders, casual_one-piece_swimsuit, hat_flower, looking_at_viewer, official_alternate_costume, solo, white_one-piece_swimsuit, black_choker, cleavage, collarbone, simple_background, blush, jewelry, white_background, braided_ponytail, cowboy_shot, one_eye_covered, thighs, black_bikini, criss-cross_halter, huge_breasts, yellow_halo
4	16	1girl, blush, casual_one-piece_swimsuit, cleavage, looking_at_viewer, official_alternate_costume, outdoors, solo, white_one-piece_swimsuit, hat_flower, black_choker, braid, collarbone, criss-cross_halter, day, jewelry, ocean, open_mouth, beach, blue_sky, bare_shoulders, bikini, one_eye_covered, smile, yellow_halo, covered_navel, cowboy_shot, cloud, yellow_headwear
5	5	1boy, 1girl, blush, hetero, solo_focus, completely_nude, huge_breasts, mosaic_censoring, paizuri, sweat, collarbone, nipples, open_mouth, penis, braid, eyes_visible_through_hair, jewelry, looking_at_viewer, pov, wet
6	5	1boy, blush, hetero, nipples, nun, solo_focus, 1girl, nude, open_mouth, sex_from_behind, white_thighhighs, eyes_visible_through_hair, garter_belt, habit, jewelry, penis, sweat, vaginal, bar_censor, collarbone, doggystyle, huge_breasts, motion_lines, navel, pussy, standing_sex, tongue_out