zhang118970/MMArt-PPR10k

---
license: apache-2.0
task_categories:
- image-text-to-text
- image-to-image
- image-to-text
- text-to-image
language:
- en
tags:
- Lightroom
- cot
- lua
- xmp
size_categories:
- 1K<n<10K
---

### MMArt-PPR10k Dataset

The `MMArt-PPR10k` is a multimodal dataset specifically created for research into the instruction-driven agentic image retouching task. It is built upon the original PPR10k dataset and offers rich, paired image data, user instructions, and information on the Lua/XMP tools used in Lightroom.

### Dataset Structure

The dataset is organized into a hierarchical folder structure. Each data sample corresponds to a specific image pair and its related files, located within a unique directory.

Here's a breakdown of the key components:

* **Unique Sample Folders**: Inside the language folder, each subdirectory (e.g., `1000_1`) represents a single data sample.
* **User Instructions**: Within each sample folder, you will find subdirectories for user instructions of varying lengths:
* `user_want_long`
* `user_want_middle`
* `user_want_short`
* **Image and Configuration Files**: Each sample folder contains the following core files:
* `before.jpg`: The original, unedited image.
* `processed.jpg`: The edited image, manipulated based on the user instructions.
* `config.lua`: The Lua configuration file used in Lightroom.
* `config.xmp`: The Xmp file, which stores metadata and editing presets for Lightroom.

### Citation
If you find MMArt useful in your research, please consider citing:
```bash
@article{jarvisart2025,
title={JarvisArt: Liberating Human Artistic Creativity via an Intelligent Photo Retouching Agent},
author={Yunlong Lin and Zixu Lin and Kunjie Lin and Jinbin Bai and Panwang Pan and Chenxin Li and Haoyu Chen and Zhongdao Wang and Xinghao Ding and Wenbo Li and Shuicheng Yan},
year={2025},
journal={arXiv preprint arXiv:2506.17612}
}
```