heig-vd-geo/URUR

---
license: mit
task_categories:
- image-segmentation
language:
- en
pretty_name: URUR
---

# URUR Dataset (Re-hosted)

**Title**: Ultra-High Resolution Segmentation with Ultra-Rich Context (URUR)

**Original authors**: Deyi Ji, Feng Zhao, Hongtao Lu, Mingyuan Tao, Jieping Ye

**Original publication**: Ji, D., Zhao, F., Lu, H., Tao, M., & Ye, J. (2023). *Ultra-High Resolution Segmentation With Ultra-Rich Context: A Novel Benchmark*. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2023, pages 23621-23630.

**Original repository**: https://github.com/jankyee/URUR

**Original licence**: MIT License

---

## Description
This dataset provides ultra-high resolution imagery and semantic segmentation masks for the benchmark described in the paper.

This is a **re-hosted** copy of the original URUR dataset in order to improve accessibility and ease of download for the research community as asked by several people on their github.

---

## Files and structure

```
/URUR-dataset/
├── images/ # High resolution input images
├── masks/ # Corresponding segmentation masks
├── class_definitions.txt # List of classes and colour palette
└── LICENSE # Original MIT licence text
```

---

## Licence and use
This dataset is distributed under the **MIT Licence** (see `LICENSE` file).
You are free to copy, modify, distribute, and use the dataset in your research or commercial projects, provided you include the copyright notice and this licence.

**Attribution**: If you use this dataset in your work, please cite the original paper as above.

# Acknowledgements

This re-hosted version is provided for ease of access.
Original dataset authors and the associated paper deserve full credit for creating the dataset and defining the benchmark.
Thank you to them for making this resource available under a permissive licence.

# Contact

If you spot any issues with the dataset, please open an issue on the original Github (https://github.com/jankyee/URUR).