Join TensorFlow at Google I/O, May 11-12 Register now

imagenet_resized

This dataset consists of the ImageNet dataset resized to fixed size. The images here are the ones provided by Chrabaszcz et. al. using the box resize method.

For downsampled ImageNet for unsupervised learning see downsampled_imagenet.

Split Examples
'train' 1,281,167
'validation' 50,000
@article{chrabaszcz2017downsampled,
  title={A downsampled variant of imagenet as an alternative to the cifar datasets},
  author={Chrabaszcz, Patryk and Loshchilov, Ilya and Hutter, Frank},
  journal={arXiv preprint arXiv:1707.08819},
  year={2017}
}

imagenet_resized/8x8 (default config)

  • Config description: Images resized to 8x8

  • Download size: 237.11 MiB

  • Feature structure:

FeaturesDict({
    'image': Image(shape=(8, 8, 3), dtype=tf.uint8),
    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=1000),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
image Image (8, 8, 3) tf.uint8
label ClassLabel tf.int64

Visualization

imagenet_resized/16x16

  • Config description: Images resized to 16x16

  • Download size: 923.34 MiB

  • Feature structure:

FeaturesDict({
    'image': Image(shape=(16, 16, 3), dtype=tf.uint8),
    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=1000),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
image Image (16, 16, 3) tf.uint8
label ClassLabel tf.int64

Visualization

imagenet_resized/32x32

  • Config description: Images resized to 32x32

  • Download size: 3.46 GiB

  • Feature structure:

FeaturesDict({
    'image': Image(shape=(32, 32, 3), dtype=tf.uint8),
    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=1000),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
image Image (32, 32, 3) tf.uint8
label ClassLabel tf.int64

Visualization

imagenet_resized/64x64

  • Config description: Images resized to 64x64

  • Download size: 13.13 GiB

  • Feature structure:

FeaturesDict({
    'image': Image(shape=(64, 64, 3), dtype=tf.uint8),
    'label': ClassLabel(shape=(), dtype=tf.int64, num_classes=1000),
})
  • Feature documentation:
Feature Class Shape Dtype Description
FeaturesDict
image Image (64, 64, 3) tf.uint8
label ClassLabel tf.int64

Visualization