The PIRM dataset consists of 200 images, which are divided into two equal sets for validation and testing. These images cover diverse contents, including people, objects, environments, flora, natural scenery, etc. Images vary in size, and are typically ~300K pixels in resolution.

This dataset was first used for evaluating the perceptual quality of super-resolution algorithms in The 2018 PIRM challenge on Perceptual Super-resolution, in conjunction with ECCV 2018.
The dataset also includes 4x down-sampled versions of all images, which were those handed out to the challenge participants.

This dataset is published under the Creative Commons Attribution-NonCommercial-ShareAlike 4.0 International License.
If you use this dataset, please cite our report:
Blau, Y., Mechrez, R., Timofte, R., Michaeli, T., Zelnik-Manor, L. "The 2018 PIRM Challenge on Perceptual Image Super-resolution." In Proceedings of the European Conference on Computer Vision Workshops (ECCVW), 2018.

Download the dataset

Baselines for SR

Here you can download the outputs (on the PIRM test set) of three baseline super-resolution algorithms:
(i) EDSR (Lim et al., CVPRW'17)
(ii) EnhanceNet (Sajjadi et al., ICCV'17)
(iii) CX (Mechrez et al., ACCV'18)

Download the baseline results

Additional results
(iv) 2018 PIRM challenge winners (3rd region, ESRGAN - Wang et al. ECCVW'18)   GitHub page
(v) 2018 PIRM challenge runner-up (3rd region, Navarrete et al. ECCVW'18)   GitHub page