PortraitCraft

# PortraitCraft
---
⚠️ This dataset is for **non-commercial academic evaluation purposes only**. Commercial use is strictly prohibited .
---

## Overview

PortraitCraft is a benchmark dataset for portrait composition understanding and portrait composition generation. It is designed to support models in learning and evaluating composition quality in portrait images. The dataset focuses on images with humans as the main subjects and includes a wide range of scenarios, such as single-person and multi-person portraits, as well as half-body and full-body compositions. It emphasizes key composition factors such as subject prominence, pose quality, image layout, and overall visual atmosphere. PortraitCraft supports two task directions: portrait composition understanding and portrait composition generation. The dataset is constructed through large-model-assisted filtering and evaluation by professional designers, ensuring high-quality samples and fine-grained composition annotations.

## Source Attribution
**Images sourced from Unsplash. Used for non-commercial academic evaluation only.**
All images are sourced from [Unsplash](https://unsplash.com/) and are subject to Unsplash's license terms. This dataset is compiled solely for academic research purposes.

---

## License and Usage Terms

This dataset is released under the **CC BY-NC 4.0 License** with additional usage restrictions:

### ✅ Permitted Use

- Non-commercial academic research
- Compositional analysis and evaluation

### ❌ Prohibited Activities

1. **Restricted Usage Scope:**
Participants are restricted to compositional analysis only. Using the original images for training commercial models or uploading them to any third-party cloud-based image processing engines is strictly prohibited.

2. **No Redistribution:**
Participants must not redistribute, resell, or repurpose this dataset for any commercial purposes without explicit authorization.

3. **Respectful Conduct:**
As this dataset contains portraits, participants must maintain academic rigor in evaluation reports and refrain from personal attacks or derogatory comments about the individuals depicted.

4. **Legal Compliance:**
This dataset is intended for academic purposes only. Participants who violate these terms shall bear full legal responsibility for their actions.

---

## Citation

If you use this dataset in your research, please cite:

```bibtex
@article{sha2026portraitcraft,
title={PortraitCraft: A Benchmark for Portrait Composition Understanding and Generation},
author={Sha, Yuyang and Lou, Zijie and Tang, Youyun and Qu, Xiaochao and Li, Haoxiang and Liu, Ting and Liu, Luoqi},
journal={arXiv preprint arXiv:2604.03611},
url={https://arxiv.org/abs/2604.03611},
year={2026}
}
```

---

## Contact

For questions, or to report misuse, please contact:
- Email: [e-mail]

---

## Disclaimer

The dataset maintainers are not responsible for any misuse of this dataset. Users must comply with all applicable laws and regulations, as well as the terms stated above. Violation of these terms may result in:

- Revocation of dataset access
- Legal action

By requesting access to this dataset, you acknowledge that you have read, understood, and agree to abide by all terms and conditions stated above.

---

# PortraitCraft
---
⚠️ 本数据集仅用于**非商业性学术评估**，严格禁止商业使用。
---

## 概述
PortraitCraft是一款面向人像构图理解与人像构图生成的基准数据集，旨在支撑模型学习与评估人像图像的构图质量。本数据集聚焦以人类为主体的图像，涵盖单人肖像、多人肖像以及半身构图、全身构图等多种场景，重点关注主体突出度、姿态质量、图像布局与整体视觉氛围等核心构图要素。PortraitCraft支持两大任务方向：人像构图理解与人像构图生成。本数据集通过大模型辅助筛选，并由专业设计师完成评估与构建，确保样本具备高质量与细粒度的构图标注。

## 来源声明
**图片来源为Unsplash，仅用于非商业性学术评估。**
所有图片均来自[Unsplash](https://unsplash.com/)，受Unsplash许可条款约束。本数据集仅为学术研究目的汇编。

---

## 许可与使用条款
本数据集采用**CC BY-NC 4.0许可协议**发布，并附加以下使用限制：

### ✅ 允许使用范围
- 非商业性学术研究
- 构图分析与评估

### ❌ 禁止行为
1. **使用范围限制：**
参与者仅可开展构图分析相关工作。严禁将原始图片用于商业模型训练，或上传至任何第三方云端图像处理引擎。
2. **禁止再分发：**
未经明确授权，参与者不得再分发、转售或为任何商业目的重新使用本数据集。
3. **合规学术操守：**
由于本数据集包含人像内容，参与者需在评估报告中保持学术严谨，不得对所描绘的个人进行人身攻击或贬损性评论。
4. **法律合规要求：**
本数据集仅用于学术目的。违反上述条款的参与者需为其行为承担全部法律责任。

---

## 引用说明
若您在研究中使用本数据集，请引用如下文献：
bibtex
@article{sha2026portraitcraft,
title={PortraitCraft: A Benchmark for Portrait Composition Understanding and Generation},
author={Sha, Yuyang and Lou, Zijie and Tang, Youyun and Qu, Xiaochao and Li, Haoxiang and Liu, Ting and Liu, Luoqi},
journal={arXiv preprint arXiv:2604.03611},
url={https://arxiv.org/abs/2604.03611},
year={2026}
}

---

## 联系方式
如有疑问或举报滥用行为，请联系：
- 邮箱：[e-mail]

---

## 免责声明
数据集维护者不对本数据集的任何滥用行为负责。用户必须遵守所有适用法律法规以及上述条款。违反条款可能导致：
- 撤销数据集访问权限
- 法律诉讼

通过申请获取本数据集，即表明您已阅读、理解并同意遵守上述所有条款与条件。