CyberHarem/hakuryu_azurlane

--- license: mit task_categories: - text-to-image tags: - art - not-for-all-audiences size_categories: - n<1K --- # Dataset of hakuryu/白龍/白龙 (Azur Lane) This is the dataset of hakuryu/白龍/白龙 (Azur Lane), containing 103 images and their tags. The core tags of this character are `horns, breasts, long_hair, asymmetrical_horns, large_breasts, earrings, white_hair, white_eyes, red_horns, grey_hair, hair_between_eyes, grey_eyes, white_horns, very_long_hair`, which are pruned in this dataset. Images are crawled from many sites (e.g. danbooru, pixiv, zerochan ...), the auto-crawling system is powered by [DeepGHS Team](https://github.com/deepghs)([huggingface organization](https://huggingface.co/deepghs)). ## List of Packages | Name | Images | Size | Download | Type | Description | |:-----------------|---------:|:-----------|:------------------------------------------------------------------------------------------------------------------|:-----------|:---------------------------------------------------------------------| | raw | 103 | 181.50 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hakuryu_azurlane/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). | | 800 | 103 | 89.69 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hakuryu_azurlane/resolve/main/dataset-800.zip) | IMG+TXT | dataset with the shorter side not exceeding 800 pixels. | | stage3-p480-800 | 259 | 195.97 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hakuryu_azurlane/resolve/main/dataset-stage3-p480-800.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. | | 1200 | 103 | 153.80 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hakuryu_azurlane/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. | | stage3-p480-1200 | 259 | 294.02 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hakuryu_azurlane/resolve/main/dataset-stage3-p480-1200.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. | ### Load Raw Dataset with Waifuc We provide raw dataset (including tagged images) for [waifuc](https://deepghs.github.io/waifuc/main/tutorials/installation/index.html) loading. If you need this, just run the following code ```python import os import zipfile from huggingface_hub import hf_hub_download from waifuc.source import LocalSource # download raw archive file zip_file = hf_hub_download( repo_id='CyberHarem/hakuryu_azurlane', repo_type='dataset', filename='dataset-raw.zip', ) # extract files to your directory dataset_dir = 'dataset_dir' os.makedirs(dataset_dir, exist_ok=True) with zipfile.ZipFile(zip_file, 'r') as zf: zf.extractall(dataset_dir) # load the dataset with waifuc source = LocalSource(dataset_dir) for item in source: print(item.image, item.meta['filename'], item.meta['tags']) ``` ## List of Clusters List of tag clustering result, maybe some outfits can be mined here. ### Raw Text Version | # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | Tags | |----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------| | 0 | 10 |  |  |  |  |  | 1girl, holding_sword, ootachi, solo, japanese_armor, looking_at_viewer, pleated_dress, skindentation, jewelry, black_gloves, coat, from_below, sakuramon, sheath, thigh_strap, cleavage, full_body, short_dress, white_background, white_sash, black_footwear, simple_background, black_dress, wide_sleeves | | 1 | 9 |  |  |  |  |  | 1girl, black_choker, cleavage, holding_sword, looking_at_viewer, ootachi, pleated_dress, solo, jewelry, coat, simple_background, huge_breasts, japanese_armor, sakuramon, white_background, black_gloves, sheath, wide_sleeves, smile, thigh_strap, white_sash | | 2 | 19 |  |  |  |  |  | 1girl, center_opening, necktie_between_breasts, official_alternate_costume, black_necktie, jewelry, looking_at_viewer, maid_headdress, solo, black_dress, multiple_horns, white_apron, puffy_short_sleeves, white_gloves, hair_on_horn, mole_on_breast, black_pantyhose, cleavage, huge_breasts, open_mouth, sharp_teeth, holding, sword_behind_back, simple_background, white_background, high_ponytail, indoors | | 3 | 11 |  |  |  |  |  | looking_at_viewer, 1girl, bare_shoulders, jewelry, mole_on_breast, solo, cleavage, dress, official_alternate_costume, smile, bangs, blush, sitting, braid, collarbone, indoors, multiple_horns, thighs, artist_name, black_footwear, floral_print, full_body, hair_on_horn, high_heels, holding, instrument, japanese_clothes, open_mouth, tassel | ### Table Version | # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | holding_sword | ootachi | solo | japanese_armor | looking_at_viewer | pleated_dress | skindentation | jewelry | black_gloves | coat | from_below | sakuramon | sheath | thigh_strap | cleavage | full_body | short_dress | white_background | white_sash | black_footwear | simple_background | black_dress | wide_sleeves | black_choker | huge_breasts | smile | center_opening | necktie_between_breasts | official_alternate_costume | black_necktie | maid_headdress | multiple_horns | white_apron | puffy_short_sleeves | white_gloves | hair_on_horn | mole_on_breast | black_pantyhose | open_mouth | sharp_teeth | holding | sword_behind_back | high_ponytail | indoors | bare_shoulders | dress | bangs | blush | sitting | braid | collarbone | thighs | artist_name | floral_print | high_heels | instrument | japanese_clothes | tassel | |----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:----------------|:----------|:-------|:-----------------|:--------------------|:----------------|:----------------|:----------|:---------------|:-------|:-------------|:------------|:---------|:--------------|:-----------|:------------|:--------------|:-------------------|:-------------|:-----------------|:--------------------|:--------------|:---------------|:---------------|:---------------|:--------|:-----------------|:--------------------------|:-----------------------------|:----------------|:-----------------|:-----------------|:--------------|:----------------------|:---------------|:---------------|:-----------------|:------------------|:-------------|:--------------|:----------|:--------------------|:----------------|:----------|:-----------------|:--------|:--------|:--------|:----------|:--------|:-------------|:---------|:--------------|:---------------|:-------------|:-------------|:-------------------|:---------| | 0 | 10 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 1 | 9 |  |  |  |  |  | X | X | X | X | X | X | X | | X | X | X | | X | X | X | X | | | X | X | | X | | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 2 | 19 |  |  |  |  |  | X | | | X | | X | | | X | | | | | | | X | | | X | | | X | X | | | X | | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | 3 | 11 |  |  |  |  |  | X | | | X | | X | | | X | | | | | | | X | X | | | | X | | | | | | X | | | X | | | X | | | | X | X | | X | | X | | | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X |

原始信息汇总

数据集概述

数据集信息

• 名称: Dataset of hakuryu/白龍/白龙 (Azur Lane)
• 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: 103张
• 核心标签: horns, breasts, long_hair, asymmetrical_horns, large_breasts, earrings, white_hair, white_eyes, red_horns, grey_hair, hair_between_eyes, grey_eyes, white_horns, very_long_hair

数据包列表

名称	图像数量	大小	类型	描述
raw	103	181.50 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	103	89.69 MiB	IMG+TXT	短边不超过800像素的数据集
stage3-p480-800	259	195.97 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素
1200	103	153.80 MiB	IMG+TXT	短边不超过1200像素的数据集
stage3-p480-1200	259	294.02 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	10		1girl, holding_sword, ootachi, solo, japanese_armor, looking_at_viewer, pleated_dress, skindentation, jewelry, black_gloves, coat, from_below, sakuramon, sheath, thigh_strap, cleavage, full_body, short_dress, white_background, white_sash, black_footwear, simple_background, black_dress, wide_sleeves
1	9		1girl, black_choker, cleavage, holding_sword, looking_at_viewer, ootachi, pleated_dress, solo, jewelry, coat, simple_background, huge_breasts, japanese_armor, sakuramon, white_background, black_gloves, sheath, wide_sleeves, smile, thigh_strap, white_sash
2	19		1girl, center_opening, necktie_between_breasts, official_alternate_costume, black_necktie, jewelry, looking_at_viewer, maid_headdress, solo, black_dress, multiple_horns, white_apron, puffy_short_sleeves, white_gloves, hair_on_horn, mole_on_breast, black_pantyhose, cleavage, huge_breasts, open_mouth, sharp_teeth, holding, sword_behind_back, simple_background, white_background, high_ponytail, indoors
3	11		looking_at_viewer, 1girl, bare_shoulders, jewelry, mole_on_breast, solo, cleavage, dress, official_alternate_costume, smile, bangs, blush, sitting, braid, collarbone, indoors, multiple_horns, thighs, artist_name, black_footwear, floral_print, full_body, hair_on_horn, high_heels, holding, instrument, japanese_clothes, open_mouth, tassel

表格版本

#	样本数量	图像示例	1girl	holding_sword	ootachi	solo	japanese_armor	looking_at_viewer	pleated_dress	skindentation	jewelry	black_gloves	coat	from_below	sakuramon	sheath	thigh_strap	cleavage	full_body	short_dress	white_background	white_sash	black_footwear	simple_background	black_dress	wide_sleeves	black_choker	huge_breasts	smile	center_opening	necktie_between_breasts	official_alternate_costume	black_necktie	maid_headdress	multiple_horns	white_apron	puffy_short_sleeves	white_gloves	hair_on_horn	mole_on_breast	black_pantyhose	open_mouth	sharp_teeth	holding	sword_behind_back	high_ponytail	indoors	bare_shoulders	dress	bangs	blush	sitting	braid	collarbone	thighs	artist_name	floral_print	high_heels	instrument	japanese_clothes	tassel
0	10		X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X
1	9		X	X	X	X	X	X	X		X	X	X		X	X	X	X			X	X		X		X	X	X	X
2	19		X			X		X			X							X			X			X	X			X		X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X
3	11		X			X		X			X							X	X				X						X			X			X				X	X		X		X			X	X	X	X	X	X	X	X	X	X	X	X	X	X	X

搜集汇总

数据集介绍

构建方式

在二次元角色数据集构建领域，针对《碧蓝航线》中的角色白龍（Hakuryu），该数据集通过自动化爬取系统从Danbooru、Pixiv、Zerochan等多个图站收集了103张高质量图像。爬取工作由DeepGHS团队开发的系统驱动，确保了数据的广泛来源与多样性。每张图像均附带了精细的标签信息，并经过剪枝处理，去除了核心标签如角、胸部、长发等冗余标记。数据集提供了多种格式的压缩包，包括原始元数据包（raw）和不同尺寸的标准化包（如800、1200像素边长限制），以及经过三阶段裁剪处理的版本（stage3），以适应不同的训练需求。

特点

该数据集的核心特点在于其精细的标签聚类与多版本适配能力。通过标签聚类分析，数据集被划分为四个主要簇，分别对应不同的服装与姿态风格，如日本盔甲装、女仆装等，这有助于研究者挖掘角色在不同装扮下的特征。此外，数据集支持通过Waifuc库直接加载原始数据，便于进行灵活的预处理。每个图像包均包含图像与对应的文本标签文件，格式统一，适合用于文本到图像生成模型的训练。数据集的规模虽小（103张图像），但通过裁剪与缩放策略，生成了多个变体版本，提升了数据的利用率。

使用方法

使用该数据集时，用户可通过Hugging Face Hub直接下载所需的压缩包。对于需要原始元数据的场景，建议采用raw版本，并借助Waifuc库的LocalSource接口加载，从而获取图像、文件名及标签信息。若需标准化尺寸的图像，可选择800或1200版本的IMG+TXT包，其短边被限制在指定像素内。对于需要局部区域的训练，推荐使用stage3系列包，它们经过三阶段裁剪，确保区域不低于480x480像素。用户可根据模型需求选择合适的版本，解压后即可直接用于训练或评估流程。

背景与挑战

背景概述

在人工智能图像生成领域，动漫角色数据集是推动文本到图像模型发展的关键资源。CyberHarem/hakuryu_azurlane数据集由DeepGHS团队于近年创建，专注于收录《碧蓝航线》中角色“白龙”（Hakuryu）的视觉素材。该团队致力于构建高质量、细粒度的动漫图像数据集，以支持角色定制化生成与多模态学习。核心研究问题在于如何通过自动爬取与标签清洗技术，从Danbooru、Pixiv等平台高效整合103张高分辨率图像，并提取如“horns”、“large_breasts”等关键属性标签，从而为模型提供精准的结构化训练数据。该数据集对动漫风格文本到图像生成领域具有示范意义，推动了角色特征可控生成的研究进展。

当前挑战

该数据集所解决的领域挑战在于动漫角色图像生成中细粒度特征对齐的难题，例如模型需从有限样本中准确复现白龙特有的不对称角、白红角组合及服饰细节。数据构建过程中面临多重挑战：其一，数据来源分散且版权复杂，需从多站点自动爬取并确保图像质量与一致性；其二，标签噪音显著，原始标签需经剪枝处理以保留核心语义，同时处理缺失或冗余标注；其三，图像尺寸与构图差异大，需通过多阶段裁剪（如stage3-p480-800）统一短边像素，并平衡数据增强与信息保留。此外，小样本量（103张）下如何避免过拟合，并利用聚类分析挖掘不同服装变体（如武士甲、女仆装）的分布规律，仍是提升泛化能力的关键瓶颈。

常用场景

经典使用场景

在二次元图像生成领域，CyberHarem/hakuryu_azurlane数据集常被用于训练和微调基于文本到图像生成模型的特定角色概念。该数据集聚焦于《碧蓝航线》中的角色白龙，包含103张经过多源爬取和标签预处理的高质量图像，其核心标签如角、长发、巨乳等特征被精心保留。研究者通常利用该数据集进行角色定制化生成，通过少量样本学习实现对该角色外观、服饰及风格的高保真再现，是风格迁移和角色概念提取任务中的经典基准数据集。

实际应用

在实际应用中，该数据集驱动了游戏角色二创、虚拟偶像内容生产及同人作品自动化生成等场景。开发者可基于其整理的标签聚类结果（如不同服装变体）快速构建特定风格的图像生成管道，降低人工绘制成本。此外，数据集提供的waifuc加载接口和多种尺寸包，使其能无缝集成到商业级图像生成工作流中，用于批量生成高质量的角色立绘、宣传物料或交互式AI绘画工具的后端模型微调。

衍生相关工作

该数据集衍生了一系列经典工作，包括基于标签聚类的角色服装分解研究、多视角角色生成模型（如通过聚类组0的持刀装甲装束学习动作与服饰关联），以及跨数据集角色迁移学习框架。研究者还利用其多版本格式开发了自适应分辨率训练策略，并催生了针对动漫角色的概念反演（Textual Inversion）和DreamBooth微调方法。这些工作进一步探索了数据增强在细粒度角色属性控制中的作用，形成了从数据构建到模型优化的完整研究链条。