downsampled_imagenet

Dataset with images of 2 resolutions (see config name for information on the resolution). It is used for density estimation and generative modeling experiments.

For resized ImageNet for supervised learning (link) see imagenet_resized.

Split Examples
'train' 1,281,149
'validation' 49,999
  • 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
@article{DBLP:journals/corr/OordKK16,
  author    = {A{"{a} }ron van den Oord and
               Nal Kalchbrenner and
               Koray Kavukcuoglu},
  title     = {Pixel Recurrent Neural Networks},
  journal   = {CoRR},
  volume    = {abs/1601.06759},
  year      = {2016},
  url       = {http://arxiv.org/abs/1601.06759},
  archivePrefix = {arXiv},
  eprint    = {1601.06759},
  timestamp = {Mon, 13 Aug 2018 16:46:29 +0200},
  biburl    = {https://dblp.org/rec/bib/journals/corr/OordKK16},
  bibsource = {dblp computer science bibliography, https://dblp.org}
}

downsampled_imagenet/32x32 (default config)

  • Config description: A dataset consisting of Train and Validation images of 32x32 resolution.

  • Download size: 3.98 GiB

  • Dataset size: 3.05 GiB

  • Figure (tfds.show_examples):

Visualization

downsampled_imagenet/64x64

  • Config description: A dataset consisting of Train and Validation images of 64x64 resolution.

  • Download size: 11.73 GiB

  • Dataset size: 10.80 GiB

  • Figure (tfds.show_examples):

Visualization