CyberHarem/kagari_pokemon

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

# Dataset of kagari/カガリ (Pokémon)

This is the dataset of kagari/カガリ (Pokémon), containing 482 images and their tags.

The core tags of this character are `purple_hair, short_hair, purple_eyes, horns, fake_horns, breasts, horned_headwear, bangs`, 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 | 482 | 484.80 MiB | [Download](https://huggingface.co/datasets/CyberHarem/kagari_pokemon/resolve/main/dataset-raw.zip) | Waifuc-Raw | Raw data with meta information (min edge aligned to 1400 if larger). |
| 800 | 482 | 300.57 MiB | [Download](https://huggingface.co/datasets/CyberHarem/kagari_pokemon/resolve/main/dataset-800.zip) | IMG+TXT | dataset with the shorter side not exceeding 800 pixels. |
| stage3-p480-800 | 1047 | 589.72 MiB | [Download](https://huggingface.co/datasets/CyberHarem/kagari_pokemon/resolve/main/dataset-stage3-p480-800.zip) | IMG+TXT | 3-stage cropped dataset with the area not less than 480x480 pixels. |
| 1200 | 482 | 441.09 MiB | [Download](https://huggingface.co/datasets/CyberHarem/kagari_pokemon/resolve/main/dataset-1200.zip) | IMG+TXT | dataset with the shorter side not exceeding 1200 pixels. |
| stage3-p480-1200 | 1047 | 800.49 MiB | [Download](https://huggingface.co/datasets/CyberHarem/kagari_pokemon/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/kagari_pokemon',
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 | 12 |  |  |  |  |  | 1girl, gloves, hoodie, looking_at_viewer, ribbed_sweater, solo, uniform, blush |
| 1 | 11 |  |  |  |  |  | 1girl, gloves, holding_poke_ball, hood_up, looking_at_viewer, solo, eyelashes, poke_ball_(basic), ribbed_dress, closed_mouth, red_dress, sweater_dress, hand_up, ribbed_sweater, white_background, boots, red_footwear |
| 2 | 16 |  |  |  |  |  | 1girl, gloves, hood_up, looking_at_viewer, eyelashes, solo, ribbed_sweater, closed_mouth, ribbed_dress, blush, upper_body, smile, sweater_dress |
| 3 | 5 |  |  |  |  |  | 1girl, boots, gloves, pokemon_(creature), hood_up, hoodie, red_footwear, ribbed_sweater, sweater_dress, closed_mouth, full_body, looking_at_viewer, ribbed_dress, smile, standing, eyelashes, red_dress |
| 4 | 20 |  |  |  |  |  | 1boy, 1girl, blush, hetero, ribbed_sweater, solo_focus, vaginal, hoodie, pussy, gloves, penis, clothed_sex, open_mouth, large_breasts, spread_legs, uniform, bar_censor, cowgirl_position, girl_on_top, lying, nipples |
| 5 | 5 |  |  |  |  |  | 1girl, blush, closed_mouth, large_breasts, long_sleeves, nipples, open_clothes, solo, hood_up, jacket, looking_at_viewer, ass_visible_through_thighs, cameltoe, simple_background, black_panties, black_thighhighs, breasts_out, cowboy_shot, gloves, heavy_breathing, hoodie, leotard, navel, piercing |

### Table Version

| # | Samples | Img-1 | Img-2 | Img-3 | Img-4 | Img-5 | 1girl | gloves | hoodie | looking_at_viewer | ribbed_sweater | solo | uniform | blush | holding_poke_ball | hood_up | eyelashes | poke_ball_(basic) | ribbed_dress | closed_mouth | red_dress | sweater_dress | hand_up | white_background | boots | red_footwear | upper_body | smile | pokemon_(creature) | full_body | standing | 1boy | hetero | solo_focus | vaginal | pussy | penis | clothed_sex | open_mouth | large_breasts | spread_legs | bar_censor | cowgirl_position | girl_on_top | lying | nipples | long_sleeves | open_clothes | jacket | ass_visible_through_thighs | cameltoe | simple_background | black_panties | black_thighhighs | breasts_out | cowboy_shot | heavy_breathing | leotard | navel | piercing |
|----:|----------:|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------------------------------|:--------|:---------|:---------|:--------------------|:-----------------|:-------|:----------|:--------|:--------------------|:----------|:------------|:--------------------|:---------------|:---------------|:------------|:----------------|:----------|:-------------------|:--------|:---------------|:-------------|:--------|:---------------------|:------------|:-----------|:-------|:---------|:-------------|:----------|:--------|:--------|:--------------|:-------------|:----------------|:--------------|:-------------|:-------------------|:--------------|:--------|:----------|:---------------|:---------------|:---------|:-----------------------------|:-----------|:--------------------|:----------------|:-------------------|:--------------|:--------------|:------------------|:----------|:--------|:-----------|
| 0 | 12 |  |  |  |  |  | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 1 | 11 |  |  |  |  |  | X | X | | X | X | X | | | X | X | X | X | X | X | X | X | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 2 | 16 |  |  |  |  |  | X | X | | X | X | X | | X | | X | X | | X | X | | X | | | | | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 3 | 5 |  |  |  |  |  | X | X | X | X | X | | | | | X | X | | X | X | X | X | | | X | X | | X | X | X | X | | | | | | | | | | | | | | | | | | | | | | | | | | | | | |
| 4 | 20 |  |  |  |  |  | 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 | 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 kagari/カガリ (Pokémon)
• 许可证: MIT
• 任务类别: text-to-image
• 标签: art, not-for-all-audiences
• 大小类别: n<1K
• 图像数量: 482
• 核心标签: purple_hair, short_hair, purple_eyes, horns, fake_horns, breasts, horned_headwear, bangs

数据集包列表

名称	图像数量	大小	类型	描述
raw	482	484.80 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）。
800	482	300.57 MiB	IMG+TXT	短边不超过800像素的数据集。
stage3-p480-800	1047	589.72 MiB	IMG+TXT	3阶段裁剪数据集，区域不小于480x480像素。
1200	482	441.09 MiB	IMG+TXT	短边不超过1200像素的数据集。
stage3-p480-1200	1047	800.49 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	12						1girl, gloves, hoodie, looking_at_viewer, ribbed_sweater, solo, uniform, blush
1	11						1girl, gloves, holding_poke_ball, hood_up, looking_at_viewer, solo, eyelashes, poke_ball_(basic), ribbed_dress, closed_mouth, red_dress, sweater_dress, hand_up, ribbed_sweater, white_background, boots, red_footwear
2	16						1girl, gloves, hood_up, looking_at_viewer, eyelashes, solo, ribbed_sweater, closed_mouth, ribbed_dress, blush, upper_body, smile, sweater_dress
3	5						1girl, boots, gloves, pokemon_(creature), hood_up, hoodie, red_footwear, ribbed_sweater, sweater_dress, closed_mouth, full_body, looking_at_viewer, ribbed_dress, smile, standing, eyelashes, red_dress
4	20						1boy, 1girl, blush, hetero, ribbed_sweater, solo_focus, vaginal, hoodie, pussy, gloves, penis, clothed_sex, open_mouth, large_breasts, spread_legs, uniform, bar_censor, cowgirl_position, girl_on_top, lying, nipples
5	5						1girl, blush, closed_mouth, large_breasts, long_sleeves, nipples, open_clothes, solo, hood_up, jacket, looking_at_viewer, ass_visible_through_thighs, cameltoe, simple_background, black_panties, black_thighhighs, breasts_out, cowboy_shot, gloves, heavy_breathing, hoodie, leotard, navel, piercing

表格版本

#	样本数量	图像1	图像2	图像3	图像4	图像5	1girl	gloves	hoodie	looking_at_viewer	ribbed_sweater	solo	uniform	blush	holding_poke_ball	hood_up	eyelashes	poke_ball_(basic)	ribbed_dress	closed_mouth	red_dress	sweater_dress	hand_up	white_background	boots	red_footwear	upper_body	smile	pokemon_(creature)	full_body	standing	1boy	hetero	solo_focus	vaginal	pussy	penis	clothed_sex	open_mouth	large_breasts	spread_legs	bar_censor	cowgirl_position	girl_on_top	lying	nipples	long_sleeves	open_clothes	jacket	ass_visible_through_thighs	cameltoe	simple_background	black_panties	black_thighhighs	breasts_out	cowboy_shot	heavy_breathing	leotard	navel	piercing
0	12						X	X	X	X	X	X	X	X
1	11						X	X		X	X	X			X	X	X	X	X	X	X	X	X	X	X	X
2	16						X	X		X	X	X		X		X	X		X	X		X					X	X
3	5						X	X	X	X	X					X	X		X	X	X	X			X	X		X	X	X	X
4	20						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	X	X	X	X	X	X

搜集汇总

数据集介绍

构建方式

在动漫与游戏角色数据集构建领域，CyberHarem/kagari_pokemon 数据集聚焦于宝可梦系列中的角色 Kagari（カガリ）。该数据集通过自动化爬虫系统从 Danbooru、Pixiv、Zerochan 等多个图像平台采集原始图像，其核心标签如紫色短发、紫色眼眸、角饰等已被系统化剔除，以保留更丰富的语义信息。数据集共包含 482 张图像，并提供了多种预处理版本：原始版保留元数据且最小边对齐至 1400 像素；缩放版将短边限制在 800 或 1200 像素以内；三阶段裁剪版则基于 480x480 像素的最小面积进行智能切割，生成 1047 张增强样本，满足不同训练需求。

使用方法

使用该数据集时，研究者可通过 Hugging Face Hub 直接下载压缩包，并借助 waifuc 框架实现高效加载。具体而言，首先利用 hf_hub_download 函数获取原始归档文件，解压至本地目录后，通过 LocalSource 接口即可迭代访问每个样本的图像及其元数据（如文件名与标签）。对于缩放版或裁剪版数据集，用户可直接解压并作为标准的 IMG+TXT 格式使用，便于集成到主流深度学习框架中。此外，聚类结果以表格形式呈现，为探索角色视觉概念提供了直观的参考，进一步简化了数据筛选与模型评估流程。

背景与挑战

背景概述

在文本到图像生成领域，高质量、精细标注的领域特定数据集是驱动模型性能跃升的关键基石。CyberHarem/kagari_pokemon数据集由DeepGHS团队于近年构建，专注于《宝可梦》系列中的角色“カガリ”（Kagari），旨在为动漫风格的角色生成任务提供标准化数据支撑。该数据集共收录482张图像，并附有详尽的标签信息，核心特征包括紫发、短发、紫瞳、角饰等，这些标签经过精心筛选以凸显角色辨识度。图像来源涵盖Danbooru、Pixiv、Zerochan等多个知名插画平台，借助自动化爬取系统高效整合。数据集的发布不仅丰富了动漫角色生成领域的素材库，更通过多尺度打包（如800px、1200px及三阶段裁剪版本）和Waifuc框架的兼容设计，为后续研究提供了灵活的预处理基准，对推动二次元风格生成模型的精细化发展具有重要参考价值。

当前挑战

该数据集所响应的核心领域挑战在于动漫角色生成任务中数据稀缺与标注不统一的问题。具体而言，现有公开数据集多聚焦于通用图像或主流角色，而特定小众角色（如Kagari）的高质量图像数量有限，且来源分散、风格各异，导致模型难以学习到一致且鲁棒的特征表达。在构建过程中，团队面临多重困难：首先，从多平台爬取图像需克服反爬机制与版权合规性，确保数据来源合法且质量可控；其次，标签体系庞杂，需从Danbooru等平台的众包标签中自动提取并统一核心属性，同时剔除冗余或噪声标签；最后，数据集规模较小（不足千张），如何通过多尺度裁剪与数据增强策略在不引入虚假关联的前提下扩充样本多样性，成为平衡过拟合与泛化能力的关键瓶颈。

常用场景

经典使用场景

在文本到图像生成领域，CyberHarem/kagari_pokemon数据集以其精细标注的角色特征（如紫色短发、紫色眼眸、角饰等）和高质量图像集合，成为微调扩散模型（如Stable Diffusion）以生成特定动漫角色形象的理想训练资源。研究者常利用其多分辨率版本（如800px和1200px）及三阶段裁剪数据集，开展角色一致性生成、属性可控合成等经典实验，从而验证模型对细粒度视觉语义的捕捉能力。

解决学术问题

该数据集有效解决了动漫角色生成中形象保真度不足与风格泛化能力弱的学术难题。通过提供涵盖多种姿态、服饰和场景的482张标准化图像及其标签聚类结果，它支撑了关于角色核心特征解耦表示、少样本概念学习以及跨域风格迁移的研究。其意义在于为可控生成模型的评估提供了客观基准，推动了文本到图像生成技术在虚拟角色创作领域的理论发展。

实际应用

在实际应用中，该数据集赋能了游戏、动画与虚拟偶像产业中的角色快速原型设计。设计师可利用其训练轻量级生成模型，自动化产出符合角色设定书的宣传图、表情包或剧情插图，显著缩短了从概念到视觉呈现的周期。此外，集成Waifuc工具的加载方式，使得个人创作者和中小团队能够低成本构建定制化角色生成管线，拓展了AI辅助内容生产的边界。

数据集最近研究

最新研究方向

在生成式人工智能与动漫文化交汇的前沿领域，基于宝可梦IP的角色数据集构建正成为文本到图像模型微调的关键基石。该数据集聚焦于卡伽莉（Kagari）这一角色，收录了482幅高质量图像及其精细化标签，并采用多尺度裁剪与聚类分析技术，为细粒度角色生成提供了结构化训练资源。当前研究方向紧密围绕二次元角色的一致性与可控性生成，通过标签修剪与图像增强策略，有效提升了模型对特定外观特征（如紫发、短角）的捕捉能力。该数据集不仅推动了动漫角色数字资产的高效复用，也为跨模态生成任务中的风格迁移与属性解耦研究提供了标准化基准，在创意内容自动化生产与虚拟角色定制化服务中展现出显著的应用潜力。