README.md

metadata

annotations_creators:
  - machine-generated
language_creators:
  - found
languages: []
licenses:
  - other-academic-use
multilinguality:
  - monolingual
pretty_name: Div2k
size_categories:
  - unknown
source_datasets:
  - original
task_categories:
  - other
task_ids:
  - other-other-image-super-resolution

Dataset Card for Div2k

Table of Contents

• Table of Contents
• Dataset Description
  • Dataset Summary
  • Supported Tasks and Leaderboards
  • Languages
• Dataset Structure
  • Data Instances
  • Data Fields
  • Data Splits
• Dataset Creation
  • Curation Rationale
  • Source Data
  • Annotations
  • Personal and Sensitive Information
• Considerations for Using the Data
  • Social Impact of Dataset
  • Discussion of Biases
  • Other Known Limitations
• Additional Information
  • Dataset Curators
  • Licensing Information
  • Citation Information
  • Contributions

Dataset Description

• Homepage: https://data.vision.ee.ethz.ch/cvl/DIV2K/
• Repository: https://huggingface.co/datasets/eugenesiow/Div2k
• Paper: http://www.vision.ee.ethz.ch/~timofter/publications/Agustsson-CVPRW-2017.pdf
• Leaderboard: https://github.com/eugenesiow/super-image#scale-x2
• Point of Contact: radu.timofte@vision.ee.ethz.ch

Dataset Summary

DIV2K is a dataset of RGB images (2K resolution high quality images) with a large diversity of contents.

The DIV2K dataset is divided into:

• train data: starting from 800 high definition high resolution images we obtain corresponding low resolution images and provide both high and low resolution images for 2, 3, and 4 downscaling factors
• validation data: 100 high definition high resolution images are used for genereting low resolution corresponding images, the low res are provided from the beginning of the challenge and are meant for the participants to get online feedback from the validation server; the high resolution images will be released when the final phase of the challenge starts.

Install with pip:

pip install datasets super-image

Evaluate a model with the super-image library:

from datasets import load_dataset
from super_image import EdsrModel
from super_image.data import EvalDataset, EvalMetrics

dataset = load_dataset('eugenesiow/Div2k', 'bicubic_x2', split='validation')
eval_dataset = EvalDataset(dataset)
model = EdsrModel.from_pretrained(input_dir, scale=scale)
EvalMetrics.evaluate(model, eval_dataset)

Supported Tasks and Leaderboards

The dataset is commonly used for training and evaluation of the image-super-resolution task.

Unofficial super-image leaderboard for:

• Scale 2
• Scale 3
• Scale 4
• Scale 8

Languages

[More Information Needed]

Dataset Structure

Data Instances

An example of train for bicubic_x2 looks as follows.

{
    "hr": "/.cache/huggingface/datasets/downloads/extracted/DIV2K_valid_HR/0801.png",
    "lr": "/.cache/huggingface/datasets/downloads/extracted/DIV2K_valid_LR_bicubic/X2/0801x2.png"
}

Data Fields

The data fields are the same among all splits.

• hr: a string to the path of the High Resolution (HR) .png image.
• lr: a string to the path of the Low Resolution (LR) .png image.

Data Splits

name	train	validation
bicubic_x2	800	100
bicubic_x3	800	100
bicubic_x4	800	100
bicubic_x8	800	100
unknown_x2	800	100
unknown_x3	800	100
unknown_x4	800	100
realistic_mild_x4	800	100
realistic_difficult_x4	800	100
realistic_wild_x4	800	100

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

Initial Data Collection and Normalization

[More Information Needed]

Who are the source language producers?

[More Information Needed]

Annotations

Annotation process

[More Information Needed]

Who are the annotators?

[More Information Needed]

Personal and Sensitive Information

All the images are collected from the Internet, and the copyright belongs to the original owners. If any of the images belongs to you and you would like it removed, please kindly inform the authors, and they will remove it from the dataset immediately.

Considerations for Using the Data

Social Impact of Dataset

[More Information Needed]

Discussion of Biases

[More Information Needed]

Other Known Limitations

[More Information Needed]

Additional Information

Dataset Curators

[More Information Needed]

Licensing Information

Please notice that this dataset is made available for academic research purpose only. All the images are collected from the Internet, and the copyright belongs to the original owners. If any of the images belongs to you and you would like it removed, please kindly inform the authors, and they will remove it from the dataset immediately.

Citation Information

@InProceedings{Agustsson_2017_CVPR_Workshops,
    author = {Agustsson, Eirikur and Timofte, Radu},
    title = {NTIRE 2017 Challenge on Single Image Super-Resolution: Dataset and Study},
    booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR) Workshops},
    url = "http://www.vision.ee.ethz.ch/~timofter/publications/Agustsson-CVPRW-2017.pdf",
    month = {July},
    year = {2017}
} 

Contributions

Thanks to @eugenesiow for adding this dataset.