clic

• Description:

CLIC is a dataset for the Challenge on Learned Image Compression 2020 lossy image compression track. These images contain a mix of the professional and mobile datasets used to train and benchmark rate-distortion performance. The dataset contains both RGB and grayscale images. This may require special handling if a grayscale image is processed as a 1 channel Tensor and a 3 channel Tensor is expected.

This dataset does NOT contain the data from the P-Frame challenge (YUV image frames).

• Additional Documentation: Explore on Papers With Code

• Homepage: https://www.compression.cc/

• Source code: tfds.datasets.clic.Builder

• Versions:

  • 1.0.0 (default): No release notes.

• Download size: 7.48 GiB

• Dataset size: 7.48 GiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'test'	428
'train'	1,633
'validation'	102

• Feature structure:

FeaturesDict({
    'image': Image(shape=(None, None, 3), dtype=uint8),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(None, None, 3)	uint8

• Supervised keys (See as_supervised doc): None

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

• Citation:

@misc{CLIC2020,
  title = {Workshop and Challenge on Learned Image Compression (CLIC2020)},
  author = {George Toderici, Wenzhe Shi, Radu Timofte, Lucas Theis,
            Johannes Balle, Eirikur Agustsson, Nick Johnston, Fabian Mentzer},
  url = {http://www.compression.cc},
  year={2020},
  organization={CVPR}
}