CyberHarem/hagikaze_kantaicollection

--- license: mit task_categories: - text-to-image tags: - art - not-for-all-audiences size_categories: - n<1K --- # Dataset of hagikaze/萩風 (Kantai Collection) This is the dataset of hagikaze/萩風 (Kantai Collection), containing 500 images and their tags. The core tags of this character are `long_hair, purple_hair, ahoge, one_side_up, brown_eyes, breasts, ribbon, red_ribbon, neck_ribbon, medium_breasts`, 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 | 500 | 464.22 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hagikaze_kantaicollection/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). | | 800 | 500 | 312.55 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hagikaze_kantaicollection/resolve/main/dataset-800.zip) | IMG+TXT | dataset with the shorter side not exceeding 800 pixels. | | stage3-p480-800 | 1166 | 657.42 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hagikaze_kantaicollection/resolve/main/dataset-stage3-p480-800.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. | | 1200 | 500 | 427.84 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hagikaze_kantaicollection/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. | | stage3-p480-1200 | 1166 | 846.92 MiB | [Download](https://huggingface.co/datasets/CyberHarem/hagikaze_kantaicollection/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/hagikaze_kantaicollection', 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 | 18 |  |  |  |  |  | 1girl, black_skirt, black_vest, pleated_skirt, short_sleeves, solo, white_gloves, white_shirt, simple_background, white_background, blouse, looking_at_viewer, school_uniform, twitter_username, smile, cowboy_shot, one-hour_drawing_challenge, blush, open_mouth | | 1 | 5 |  |  |  |  |  | 1girl, black_skirt, black_vest, blouse, full_body, grey_socks, kneehighs, pleated_skirt, short_sleeves, simple_background, solo, white_background, white_gloves, white_shirt, school_uniform, black_footwear, chibi, loafers, looking_at_viewer, smile, standing | | 2 | 6 |  |  |  |  |  | 1girl, black_skirt, black_vest, blouse, pleated_skirt, purple_panties, short_sleeves, solo, white_shirt, white_gloves, lifted_by_self, skirt_lift, dress_shirt, sitting | | 3 | 9 |  |  |  |  |  | 1girl, black_vest, blouse, gradient_background, short_sleeves, solo, upper_body, white_gloves, white_shirt, looking_at_viewer, open_mouth, dated, one-hour_drawing_challenge, red_background, blush, school_uniform | | 4 | 12 |  |  |  |  |  | 1girl, navel, simple_background, solo, underwear_only, blush, cleavage, collarbone, looking_at_viewer, white_background, purple_panties, twitter_username, purple_bra, cowboy_shot, large_breasts, open_mouth, one-hour_drawing_challenge, smile, bare_shoulders, groin, white_gloves, yellow_eyes | | 5 | 5 |  |  |  |  |  | 1girl, blush, collarbone, large_breasts, looking_at_viewer, solo, cleavage, smile, dated, purple_bikini, purple_eyes, alternate_costume, cowboy_shot, navel, one-hour_drawing_challenge, simple_background, twitter_username, white_background | | 6 | 11 |  |  |  |  |  | 1girl, black_dress, simple_background, solo, enmaided, looking_at_viewer, white_apron, blush, maid_apron, frilled_apron, white_background, maid_headdress, open_mouth, short_sleeves, smile, white_gloves, cleavage, hair_between_eyes, one-hour_drawing_challenge, thighhighs, twitter_username | | 7 | 9 |  |  |  |  |  | detached_collar, playboy_bunny, rabbit_ears, 1girl, fake_animal_ears, solo, looking_at_viewer, strapless_leotard, wrist_cuffs, blush, cleavage, open_mouth, smile, alternate_costume, black_leotard, bowtie, fishnet_pantyhose, large_breasts, simple_background, white_background, bare_shoulders, covered_navel, rabbit_tail | | 8 | 7 |  |  |  |  |  | 1girl, alternate_costume, obi, solo, yukata, looking_at_viewer, smile, alternate_hairstyle, blue_kimono, ponytail, simple_background, white_background, hairclip, open_mouth, bag, blush, dated, full_body, long_sleeves, sandals, twitter_username, wide_sleeves, yellow_eyes | ### Table Version | # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | black_skirt | black_vest | pleated_skirt | short_sleeves | solo | white_gloves | white_shirt | simple_background | white_background | blouse | looking_at_viewer | school_uniform | twitter_username | smile | cowboy_shot | one-hour_drawing_challenge | blush | open_mouth | full_body | grey_socks | kneehighs | black_footwear | chibi | loafers | standing | purple_panties | lifted_by_self | skirt_lift | dress_shirt | sitting | gradient_background | upper_body | dated | red_background | navel | underwear_only | cleavage | collarbone | purple_bra | large_breasts | bare_shoulders | groin | yellow_eyes | purple_bikini | purple_eyes | alternate_costume | black_dress | enmaided | white_apron | maid_apron | frilled_apron | maid_headdress | hair_between_eyes | thighhighs | detached_collar | playboy_bunny | rabbit_ears | fake_animal_ears | strapless_leotard | wrist_cuffs | black_leotard | bowtie | fishnet_pantyhose | covered_navel | rabbit_tail | obi | yukata | alternate_hairstyle | blue_kimono | ponytail | hairclip | bag | long_sleeves | sandals | wide_sleeves | |----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:--------------|:-------------|:----------------|:----------------|:-------|:---------------|:--------------|:--------------------|:-------------------|:---------|:--------------------|:-----------------|:-------------------|:--------|:--------------|:-----------------------------|:--------|:-------------|:------------|:-------------|:------------|:-----------------|:--------|:----------|:-----------|:-----------------|:-----------------|:-------------|:--------------|:----------|:----------------------|:-------------|:--------|:-----------------|:--------|:-----------------|:-----------|:-------------|:-------------|:----------------|:-----------------|:--------|:--------------|:----------------|:--------------|:--------------------|:--------------|:-----------|:--------------|:-------------|:----------------|:-----------------|:--------------------|:-------------|:------------------|:----------------|:--------------|:-------------------|:--------------------|:--------------|:----------------|:---------|:--------------------|:----------------|:--------------|:------|:---------|:----------------------|:--------------|:-----------|:-----------|:------|:---------------|:----------|:---------------| | 0 | 18 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 1 | 5 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | X | | X | | | | | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 2 | 6 |  |  |  |  |  | X | X | X | X | X | X | X | X | | | X | | | | | | | | | | | | | | | | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 3 | 9 |  |  |  |  |  | X | | X | | X | X | X | X | | | X | X | X | | | | X | X | X | | | | | | | | | | | | | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 4 | 12 |  |  |  |  |  | X | | | | | X | X | | X | X | | X | | X | X | X | X | X | X | | | | | | | | X | | | | | | | | | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 5 | 5 |  |  |  |  |  | X | | | | | X | | | X | X | | X | | X | X | X | X | X | | | | | | | | | | | | | | | | X | | X | | X | X | | X | | | | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | 6 | 11 |  |  |  |  |  | X | | | | X | X | X | | X | X | | X | | X | X | | X | X | X | | | | | | | | | | | | | | | | | | | X | | | | | | | | | | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | 7 | 9 |  |  |  |  |  | X | | | | | X | | | X | X | | X | | | X | | | X | X | | | | | | | | | | | | | | | | | | | X | | | X | X | | | | | X | | | | | | | | | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | 8 | 7 |  |  |  |  |  | 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

萩風 (Kantai Collection) 数据集

数据集概述

该数据集包含萩風 (Kantai Collection) 的500张图像及其标签。核心标签包括 long_hair, purple_hair, ahoge, one_side_up, brown_eyes, breasts, ribbon, red_ribbon, neck_ribbon, medium_breasts。

数据集包列表

名称	图像数量	大小	类型	描述
raw	500	464.22 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	500	312.55 MiB	IMG+TXT	短边不超过800像素的数据集。
stage3-p480-800	1166	657.42 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素。
1200	500	427.84 MiB	IMG+TXT	短边不超过1200像素的数据集。
stage3-p480-1200	1166	846.92 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1	图像2	图像3	图像4	图像5	标签
0	18	样本0	样本1	样本2	样本3	样本4	1girl, black_skirt, black_vest, pleated_skirt, short_sleeves, solo, white_gloves, white_shirt, simple_background, white_background, blouse, looking_at_viewer, school_uniform, twitter_username, smile, cowboy_shot, one-hour_drawing_challenge, blush, open_mouth
1	5	样本0	样本1	样本2	样本3	样本4	1girl, black_skirt, black_vest, blouse, full_body, grey_socks, kneehighs, pleated_skirt, short_sleeves, simple_background, solo, white_background, white_gloves, white_shirt, school_uniform, black_footwear, chibi, loafers, looking_at_viewer, smile, standing
2	6	样本0	样本1	样本2	样本3	样本4	1girl, black_skirt, black_vest, blouse, pleated_skirt, purple_panties, short_sleeves, solo, white_shirt, white_gloves, lifted_by_self, skirt_lift, dress_shirt, sitting
3	9	样本0	样本1	样本2	样本3	样本4	1girl, black_vest, blouse, gradient_background, short_sleeves, solo, upper_body, white_gloves, white_shirt, looking_at_viewer, open_mouth, dated, one-hour_drawing_challenge, red_background, blush, school_uniform
4	12	样本0	样本1	样本2	样本3	样本4	1girl, navel, simple_background, solo, underwear_only, blush, cleavage, collarbone, looking_at_viewer, white_background, purple_panties, twitter_username, purple_bra, cowboy_shot, large_breasts, open_mouth, one-hour_drawing_challenge, smile, bare_shoulders, groin, white_gloves, yellow_eyes
5	5	样本0	样本1	样本2	样本3	样本4	1girl, blush, collarbone, large_breasts, looking_at_viewer, solo, cleavage, smile, dated, purple_bikini, purple_eyes, alternate_costume, cowboy_shot, navel, one-hour_drawing_challenge, simple_background, twitter_username, white_background
6	11	样本0	样本1	样本2	样本3	样本4	1girl, black_dress, simple_background, solo, enmaided, looking_at_viewer, white_apron, blush, maid_apron, frilled_apron, white_background, maid_headdress, open_mouth, short_sleeves, smile, white_gloves, cleavage, hair_between_eyes, one-hour_drawing_challenge, thighhighs, twitter_username
7	9	样本0	样本1	样本2	样本3	样本4	detached_collar, playboy_bunny, rabbit_ears, 1girl, fake_animal_ears, solo, looking_at_viewer, strapless_leotard, wrist_cuffs, blush, cleavage, open_mouth, smile, alternate_costume, black_leotard, bowtie, fishnet_pantyhose, large_breasts, simple_background, white_background, bare_shoulders, covered_navel, rabbit_tail
8	7	样本0	样本1	样本2	样本3	样本4	1girl, alternate_costume, obi, solo, yukata, looking_at_viewer, smile, alternate_hairstyle, blue_kimono, ponytail, simple_background, white_background, hairclip, open_mouth, bag, blush, dated, full_body, long_sleeves, sandals, twitter_username, wide_sleeves, yellow_eyes

表格版本

#	样本数量	图像1	图像2	图像3	图像4	图像5	1girl	black_skirt	black_vest	pleated_skirt	short_sleeves	solo	white_gloves	white_shirt	simple_background	white_background	blouse	looking_at_viewer	school_uniform	twitter_username	smile	cowboy_shot	one-hour_drawing_challenge	blush	open_mouth	full_body	grey_socks	kneehighs	black_footwear	chibi	loafers	standing	purple_panties	lifted_by_self	skirt_lift	dress_shirt	sitting	gradient_background	upper_body	dated	red_background	navel	underwear_only	cleavage	collarbone	purple_bra	large_breasts	bare_shoulders	groin	yellow_eyes	purple_bikini	purple_eyes	alternate_costume	black_dress	enmaided	white_apron	maid_apron	frilled_apron	maid_headdress	hair_between_eyes	thighhighs	detached_collar	playboy_bunny	rabbit_ears	fake_animal_ears	strapless_leotard	wrist_cuffs	black_leotard	bowtie	fishnet_pantyhose	covered_navel	rabbit_tail	obi	yukata	alternate_hairstyle	blue_kimono	ponytail	hairclip	bag	long_sleeves	sandals	wide_sleeves
0	18	样本0	样本1	样本2	样本3	样本4	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X	X
1	5	样本0	样本1	样本2	样本3	样本4	X	X	X	X	X	X	X	X	X	X	X	X	X		X					X	X	X	X	X	X	X
2	6	样本0	样本1	样本2	样本3	样本4	X	X	X	X	X	X	X	X			X																X	X	X	X	X
3	9	样本0	样本1	样本2	样本3	样本4	X		X		X	X	X	X			X	X	X				X	X	X													X	X	X	X
4	12	样本0	样本1	样本2	样本3	样本4	X					X	X		X	X		X		X	X	X	X	X	X								X									X	X	X	X	X	X	X	X	X
5	5	样本0	样本1	样本2	样本3	样本4	X					X			X	X		X		X	X	X

搜集汇总

数据集介绍

构建方式

该数据集聚焦于《舰队Collection》中的角色萩風，通过自动化爬取系统从Danbooru、Pixiv、Zerochan等多个图像平台采集了500幅高质量图像及其对应的标注标签。爬取工作由DeepGHS团队自主开发的工具链驱动，确保了数据来源的多样性与采集效率。在数据构建过程中，团队对原始图像进行了尺寸统一处理，提供了多种分辨率版本（如短边不超过800像素或1200像素），并生成了经过三阶段裁剪的增强数据集，以适配不同训练需求。所有图像均保留了元信息，并依据角色核心特征（如长发、紫发、呆毛等）对标签进行了精简与优化。

特点

数据集的显著特点在于其层次化的组织架构与丰富的衍生版本。除了原始未处理的图像档案外，还提供了面向模型训练的标准化IMG+TXT格式压缩包，极大便利了文本到图像生成任务的开展。尤为突出的是，数据集内嵌了基于标签聚类的分析结果，将图像按服饰风格（如水手服、兔女郎装、和服等）划分为多个子簇，并附带了可视化样本与详细的标签矩阵。这种结构不仅揭示了角色在不同装扮下的视觉规律，也为风格迁移、概念解耦等高级研究提供了天然的数据基础。

使用方法

使用者可通过HuggingFace Hub直接下载不同规格的数据包。对于需要灵活处理原始标注信息的研究者，推荐采用waifuc框架加载原始档案，通过简单的Python脚本即可遍历图像、文件名及关联标签。具体实现上，首先利用huggingface_hub库下载dataset-raw.zip压缩包并解压至本地目录，随后实例化waifuc的LocalSource对象指向该目录，最终通过迭代器逐项获取图像与元数据。这一流程简化了数据加载的复杂度，使研究者能快速将数据整合至自定义训练管线中。

背景与挑战

背景概述

在二次元文化蓬勃发展的当下，动漫角色数据集成为文本生成图像等视觉创作任务的重要基石。CyberHarem/hagikaze_kantaicollection数据集由DeepGHS团队于近期构建，聚焦于《舰队Collection》中的经典角色“萩風”。该数据集的核心研究问题在于为特定动漫角色提供高质量、多姿态且标注完善的图像资源，以支持角色驱动的图像生成与风格迁移研究。通过整合来自Danbooru、Pixiv、Zerochan等主流插画社区的500张图像，并辅以自动化的标签体系，该数据集为二次元视觉模型训练提供了标准化素材，对推动动漫角色图像生成领域的精细化发展具有显著的示范意义。

当前挑战

该数据集面临的核心挑战在于动漫图像领域特有的标注歧义性与风格多样性问题。首先，同一角色在不同画师笔下可能呈现差异化的视觉特征（如发型、服饰细节），导致标签系统难以完全覆盖所有变体，进而影响模型对角色核心特征的稳定学习。其次，数据采集过程中自动爬取系统虽提升了效率，但来自多源的图像在分辨率、构图、背景复杂度上高度不均，需通过多阶段裁剪与尺寸标准化（如480x480像素区域限制）来缓解，这一过程易引入裁剪偏差或丢失关键语义信息。此外，标签的稀疏性与冗余性并存，部分高频标签（如“1girl”“solo”）对角色辨识贡献有限，而细粒度属性（如“purple_panties”“alternate_costume”）的标注一致性仍需人工校验，这对数据集的可用性构成了潜在制约。

常用场景

经典使用场景

在二次元角色生成与风格迁移领域，CyberHarem/hagikaze_kantaicollection 数据集为文本到图像生成模型提供了高质量的微调素材。该数据集收录了来自《舰队Collection》中角色萩風的500张图像及详细标签，涵盖多种画风与姿势，经典用途在于训练模型精准捕捉角色核心特征（如紫色长发、呆毛、红色缎带等），并生成符合角色设定的新图像。研究者常利用此数据集进行角色一致性生成、风格适配以及标签驱动的图像合成实验。

解决学术问题

该数据集有效解决了动漫角色生成中身份特征保持与多样性平衡的学术难题。通过提供多来源、多尺寸的标注图像，它支撑了条件生成对抗网络（cGANs）和扩散模型（如Stable Diffusion）在角色定制化上的研究，推动了少样本学习、标签噪声处理以及跨模态对齐等方向的发展。其意义在于为动漫内容自动生成、数字娱乐资产创作提供了可复现的基准，促进了生成式模型在特定领域泛化能力的提升。

衍生相关工作

基于此数据集，衍生出了一系列经典工作，包括角色特定超分辨率模型、基于标签的细粒度图像编辑方法以及跨角色风格迁移框架。例如，研究者利用其标签聚类结果训练了面向服饰变化的属性解耦网络，并开发了多阶段裁剪策略以提升训练效率。这些工作进一步拓展了数据集在可控图像生成、少样本角色识别以及动漫数据集标准化构建中的应用边界。