Datasets:

Modalities:
Image
ArXiv:
Libraries:
Datasets
License:
StableSR-TestSets / README.md
Iceclear's picture
Update README.md
a4a8b2a
---
license: other
license_name: ntu-slab-license
license_link: https://github.com/IceClear/StableSR/blob/main/LICENSE.txt
task_categories:
- image-to-image
---
# StableSR TestSets Card
These test sets are used associated with the StableSR, available [here](https://github.com/IceClear/StableSR).
## Data Details
- **Developed by:** Jianyi Wang
- **Data type:** Synthetic and real-world test sets for image super-resolution
- **License:** [S-Lab License 1.0](https://github.com/IceClear/StableSR/blob/main/LICENSE.txt)
- **Data Description:** The test sets are used to reproduce the metric results shown in [Paper](https://arxiv.org/abs/2305.07015).
- **Resources for more information:** [GitHub Repository](https://github.com/IceClear/StableSR).
- **Cite as:**
@InProceedings{wang2023exploiting,
author = {Wang, Jianyi and Yue, Zongsheng and Zhou, Shangchen and Chan, Kelvin CK and Loy, Chen Change},
title = {Exploiting Diffusion Prior for Real-World Image Super-Resolution},
booktitle = {arXiv preprint arXiv:2305.07015},
year = {2023},
}
# Uses
Please refer to [S-Lab License 1.0](https://github.com/IceClear/StableSR/blob/main/LICENSE.txt)
We currently provide the following test sets:
- DIV2K_Val: 3000 synthetic data pairs on the validation of [DIV2K](https://data.vision.ee.ethz.ch/cvl/DIV2K/) generated used the same degradation used for training StableSR.
- RealSR Val: Center-cropped data pairs on [RealSRv3](https://github.com/csjcai/RealSR).
- DRealSR Val: Center-cropped data pairs on [DRealSR](https://github.com/xiezw5/Component-Divide-and-Conquer-for-Real-World-Image-Super-Resolution).
- DPED Val: Center-cropped LQ-only data on [DPED](https://github.com/aiff22/DPED).
## Evaluation Results
See [Paper](https://arxiv.org/abs/2305.07015) for details.