CyberHarem/chiyoda_kantaicollection

---
license: mit
task_categories:
- text-to-image
tags:
- art
- not-for-all-audiences
size_categories:
- n<1K
---

# Dataset of chiyoda/千代田/千代田 (Kantai Collection)

This is the dataset of chiyoda/千代田/千代田 (Kantai Collection), containing 262 images and their tags.

The core tags of this character are `brown_hair, brown_eyes, short_hair, breasts, large_breasts, headband`, 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 | 262 | 194.84 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chiyoda_kantaicollection/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 800 | 262 | 142.48 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chiyoda_kantaicollection/resolve/main/dataset-800.zip) | IMG+TXT | dataset with the shorter side not exceeding 800 pixels. |
| stage3-p480-800 | 543 | 275.36 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chiyoda_kantaicollection/resolve/main/dataset-stage3-p480-800.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. |
| 1200 | 262 | 183.59 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chiyoda_kantaicollection/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 543 | 339.23 MiB | [Download](https://huggingface.co/datasets/CyberHarem/chiyoda_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/chiyoda_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 | 35 |  |  |  |  |  | 1girl, solo, jacket, blouse, red_hakama, open_mouth, hakama_short_skirt, thighhighs, looking_at_viewer, smile, blush, white_background |
| 1 | 24 |  |  |  |  |  | 1girl, solo, blush, looking_at_viewer, open_mouth |
| 2 | 7 |  |  |  |  |  | 1girl, blush, looking_at_viewer, navel, one-hour_drawing_challenge, simple_background, solo, twitter_username, white_background, cleavage, collarbone, stomach, bare_shoulders, cropped_legs, alternate_costume, bangs, black_bikini, cowboy_shot, medium_hair, skindentation, front-tie_bikini_top, side-tie_bikini_bottom, spoken_squiggle |
| 3 | 6 |  |  |  |  |  | 1girl, blue_one-piece_swimsuit, looking_at_viewer, solo, collarbone, dated, simple_background, twitter_username, white_background, competition_swimsuit, covered_navel, cowboy_shot, one-hour_drawing_challenge, cleavage, highleg_swimsuit, sitting |
| 4 | 11 |  |  |  |  |  | blush, hetero, solo_focus, 1boy, 1girl, nipples, paizuri, looking_at_viewer, sweat, huge_breasts, penis, bar_censor, open_mouth |
| 5 | 7 |  |  |  |  |  | 1boy, 1girl, blush, hetero, nipples, solo_focus, vaginal, nude, open_mouth, girl_on_top, huge_breasts, navel, penis, bar_censor, cowgirl_position, cum_in_pussy, heart, large_areolae, sex_from_behind, sweat |
| 6 | 12 |  |  |  |  |  | 1girl, detached_collar, fake_animal_ears, looking_at_viewer, playboy_bunny, rabbit_ears, strapless_leotard, solo, simple_background, cleavage, white_background, wrist_cuffs, blush, bowtie, rabbit_tail, black_pantyhose, alternate_costume, black_leotard, cowboy_shot, fake_tail, twitter_username |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | solo | jacket | blouse | red_hakama | open_mouth | hakama_short_skirt | thighhighs | looking_at_viewer | smile | blush | white_background | navel | one-hour_drawing_challenge | simple_background | twitter_username | cleavage | collarbone | stomach | bare_shoulders | cropped_legs | alternate_costume | bangs | black_bikini | cowboy_shot | medium_hair | skindentation | front-tie_bikini_top | side-tie_bikini_bottom | spoken_squiggle | blue_one-piece_swimsuit | dated | competition_swimsuit | covered_navel | highleg_swimsuit | sitting | hetero | solo_focus | 1boy | nipples | paizuri | sweat | huge_breasts | penis | bar_censor | vaginal | nude | girl_on_top | cowgirl_position | cum_in_pussy | heart | large_areolae | sex_from_behind | detached_collar | fake_animal_ears | playboy_bunny | rabbit_ears | strapless_leotard | wrist_cuffs | bowtie | rabbit_tail | black_pantyhose | black_leotard | fake_tail |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:-------|:---------|:---------|:-------------|:-------------|:---------------------|:-------------|:--------------------|:--------|:--------|:-------------------|:--------|:-----------------------------|:--------------------|:-------------------|:-----------|:-------------|:----------|:-----------------|:---------------|:--------------------|:--------|:---------------|:--------------|:--------------|:----------------|:-----------------------|:-------------------------|:------------------|:--------------------------|:--------|:-----------------------|:----------------|:-------------------|:----------|:---------|:-------------|:-------|:----------|:----------|:--------|:---------------|:--------|:-------------|:----------|:-------|:--------------|:-------------------|:---------------|:--------|:----------------|:------------------|:------------------|:-------------------|:----------------|:--------------|:--------------------|:--------------|:---------|:--------------|:------------------|:----------------|:------------|
| 0 | 35 |  |  |  |  |  | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 24 |  |  |  |  |  | X | X | | | | X | | | X | | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 7 |  |  |  |  |  | X | X | | | | | | | X | | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 6 |  |  |  |  |  | X | X | | | | | | | X | | | X | | X | X | X | X | X | | | | | | | X | | | | | | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 11 |  |  |  |  |  | X | | | | | X | | | X | | X | | | | | | | | | | | | | | | | | | | | | | | | | | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | |
| 5 | 7 |  |  |  |  |  | X | | | | | X | | | | | X | | X | | | | | | | | | | | | | | | | | | | | | | | | X | X | X | X | | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | |
| 6 | 12 |  |  |  |  |  | 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: chiyoda/千代田/千代田 (Kantai Collection)
• 描述: 包含262张图片及其标签的数据集。
• 核心标签: brown_hair, brown_eyes, short_hair, breasts, large_breasts, headband

数据集包列表

名称	图片数量	大小	类型	描述
raw	262	194.84 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	262	142.48 MiB	IMG+TXT	短边不超过800像素的数据集。
stage3-p480-800	543	275.36 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素。
1200	262	183.59 MiB	IMG+TXT	短边不超过1200像素的数据集。
stage3-p480-1200	543	339.23 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	图片示例	标签
0	35		1girl, solo, jacket, blouse, red_hakama, open_mouth, hakama_short_skirt, thighhighs, looking_at_viewer, smile, blush, white_background
1	24		1girl, solo, blush, looking_at_viewer, open_mouth
2	7		1girl, blush, looking_at_viewer, navel, one-hour_drawing_challenge, simple_background, solo, twitter_username, white_background, cleavage, collarbone, stomach, bare_shoulders, cropped_legs, alternate_costume, bangs, black_bikini, cowboy_shot, medium_hair, skindentation, front-tie_bikini_top, side-tie_bikini_bottom, spoken_squiggle
3	6		1girl, blue_one-piece_swimsuit, looking_at_viewer, solo, collarbone, dated, simple_background, twitter_username, white_background, competition_swimsuit, covered_navel, cowboy_shot, one-hour_drawing_challenge, cleavage, highleg_swimsuit, sitting
4	11		blush, hetero, solo_focus, 1boy, 1girl, nipples, paizuri, looking_at_viewer, sweat, huge_breasts, penis, bar_censor, open_mouth
5	7		1boy, 1girl, blush, hetero, nipples, solo_focus, vaginal, nude, open_mouth, girl_on_top, huge_breasts, navel, penis, bar_censor, cowgirl_position, cum_in_pussy, heart, large_areolae, sex_from_behind, sweat
6	12		1girl, detached_collar, fake_animal_ears, looking_at_viewer, playboy_bunny, rabbit_ears, strapless_leotard, solo, simple_background, cleavage, white_background, wrist_cuffs, blush, bowtie, rabbit_tail, black_pantyhose, alternate_costume, black_leotard, cowboy_shot, fake_tail, twitter_username

表格版本

#	样本数量	图片示例	1girl	solo	jacket	blouse	red_hakama	open_mouth	hakama_short_skirt	thighhighs	looking_at_viewer	smile	blush	white_background	navel	one-hour_drawing_challenge	simple_background	twitter_username	cleavage	collarbone	stomach	bare_shoulders	cropped_legs	alternate_costume	bangs	black_bikini	cowboy_shot	medium_hair	skindentation	front-tie_bikini_top	side-tie_bikini_bottom	spoken_squiggle	blue_one-piece_swimsuit	dated	competition_swimsuit	covered_navel	highleg_swimsuit	sitting	hetero	solo_focus	1boy	nipples	paizuri	sweat	huge_breasts	penis	bar_censor	vaginal	nude	girl_on_top	cowgirl_position	cum_in_pussy	heart	large_areolae	sex_from_behind	detached_collar	fake_animal_ears	playboy_bunny	rabbit_ears	strapless_leotard	wrist_cuffs	bowtie	rabbit_tail	black_pantyhose	black_leotard	fake_tail
0	35		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，该数据集聚焦于《舰队Collection》中的角色千代田，共收录262幅图像及其标签。数据采集过程依托DeepGHS团队开发的自动化爬取系统，从Danbooru、Pixiv、Zerochan等多个知名插画平台广泛获取素材。为确保数据质量，数据集对角色核心标签（如棕色头发、棕色眼睛、短发、胸部、巨乳、发带）进行了系统性剪枝处理。数据以多种分辨率形式发布，包括原始数据包（最小边对齐至1400像素）、800与1200像素限制版本，以及经过三阶段裁剪的480像素以上区域版本，满足不同训练需求。

特点

该数据集的一大特色在于其精细的标签聚类分析，通过自动聚类算法将图像划分为7个视觉相似群组，每组附带详细的标签集合。聚类结果揭示了角色在不同场景下的多样性表现，从日常装束（如红色袴裙、夹克）到特殊装扮（如泳装、兔女郎服饰）乃至成人向内容均有涵盖。每个聚类均提供样本预览和完整标签列表，便于研究者快速理解数据分布。数据集还包含元数据信息，支持图像与标签的关联检索，为多模态学习提供了结构化基础。

使用方法

数据集提供了灵活的使用方式。研究者可通过HuggingFace Hub直接下载压缩包，其中原始数据包兼容Waifuc框架，用户只需下载并解压后，利用LocalSource即可加载图像及其元数据（包括文件名和标签）。对于需要特定分辨率的任务，可选用800或1200像素版本；若需增强数据多样性，三阶段裁剪版本通过区域筛选生成了543个样本，适合细粒度特征学习。标签聚类结果以表格形式呈现，可直接用于分析角色常见属性组合或训练条件生成模型。

背景与挑战

背景概述

在生成式人工智能与动漫文化交相辉映的时代，高质量、细粒度的角色图像数据集成为推动文本到图像生成模型发展的核心基石。CyberHarem/chiyoda_kantaicollection 数据集由 DeepGHS 团队于近期构建并发布，聚焦于《舰队Collection》中的经典角色“千代田”。该数据集旨在通过整合来自 Danbooru、Pixiv 等多元平台的 262 幅图像及其详尽标签，为角色一致性生成与细粒度风格迁移研究提供标准化资源。其影响力在于，它填补了特定二次元角色在公开数据集中的稀缺性，为 LoRA 微调、概念定制等前沿技术提供了可复现的实验基准，进而推动了动漫领域视觉生成任务从泛化到专精的跨越。

当前挑战

当前数据集面临的核心挑战体现在三个层面：首先，在领域问题层面，尽管文本到图像生成模型已取得长足进步，但针对特定动漫角色（如千代田）的精细特征（如发型、服饰细节）保持语义一致性与视觉连贯性仍属难题，现有模型常出现特征混淆或风格漂移；其次，在构建过程中，从多个异构网站自动爬取图像时，面临版权归属模糊、图像质量参差不齐以及标签噪声（如缺失或冗余标签）的干扰，需要依赖 Waifuc 框架进行多阶段清洗与裁剪；最后，数据规模仅含 262 张原始图像，样本量偏小易导致过拟合，且聚类分析揭示的服装变体（如泳装、兔女郎装）间分布不均衡，进一步加剧了模型泛化的挑战。

常用场景

经典使用场景

在生成式人工智能蓬勃发展的时代背景下，针对特定动漫角色的高质量图像生成模型训练，对数据集的精细度与多样性提出了严苛要求。CyberHarem/chiyoda_kantaicollection 数据集专为《舰队Collection》中千代田这一角色设计，收录了262张源自Danbooru、Pixiv等知名社区的高清图像，并附有精准的标签注解。其经典使用场景在于作为文生图（text-to-image）模型的微调训练语料，通过提供角色核心特征标签（如棕色长发、大胸、发箍等），使模型能够精准捕捉并再现千代田的独特视觉风格，从而在角色定制化生成任务中达到高度保真与风格一致性。

实际应用

在实际应用层面，该数据集为动漫创作、游戏开发与虚拟偶像产业提供了高效的技术支撑。创作者可基于此数据集训练定制化的图像生成模型，快速产出符合角色设定的宣传图、同人插画或动态立绘，大幅缩短设计周期并降低人工成本。例如，游戏工作室可利用其对千代田的泳装、兔女郎等变体服饰进行即时生成，用于角色皮肤设计或活动素材制作。此外，数据集提供的标签聚类结果（如红袴、比基尼等）可直接服务于自动化图像分类与检索系统，提升数字资产管理效率。

衍生相关工作

该数据集的发布催生了一系列衍生研究工作，尤其在角色识别与多模态学习领域产生了深远影响。基于其标签体系与聚类分析，研究者开发了针对二次元角色的细粒度属性分类器，实现了从发型、服饰到姿态的精准识别。同时，数据集与Waifuc框架的深度集成，推动了自动化数据爬取与预处理流程的标准化，衍生出如“CyberHarem”系列数据集构建方法论，为其他动漫角色的数据积累提供了可复现的范式。此外，在少样本学习与迁移学习任务中，该数据集常被用作基准测试集，用以评估模型在稀缺数据场景下的特征提取与生成能力。