emnist

• Description:

The EMNIST dataset is a set of handwritten character digits derived from the NIST Special Database 19 and converted to a 28x28 pixel image format and dataset structure that directly matches the MNIST dataset.

Note: Like the original EMNIST data, images provided here are inverted horizontally and rotated 90 anti-clockwise. You can use tf.transpose within ds.map to convert the images to a human-friendlier format.

• Additional Documentation: Explore on Papers With Code

• Homepage: https://www.nist.gov/itl/products-and-services/emnist-dataset

• Source code: tfds.image_classification.EMNIST

• Versions:

  • 3.0.0: New split API (https://tensorflow.org/datasets/splits)
  • 3.1.0 (default): Updated broken download URL

• Download size: 535.73 MiB

• Supervised keys (See as_supervised doc): ('image', 'label')

• Citation:

@article{cohen_afshar_tapson_schaik_2017,
    title={EMNIST: Extending MNIST to handwritten letters},
    DOI={10.1109/ijcnn.2017.7966217},
    journal={2017 International Joint Conference on Neural Networks (IJCNN)},
    author={Cohen, Gregory and Afshar, Saeed and Tapson, Jonathan and Schaik, Andre Van},
    year={2017}
}

emnist/byclass (default config)

• Config description: EMNIST ByClass

• Dataset size: 349.16 MiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'test'	116,323
'train'	697,932

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=62),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

emnist/bymerge

• Config description: EMNIST ByMerge

• Dataset size: 349.16 MiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'test'	116,323
'train'	697,932

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=47),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

emnist/balanced

• Config description: EMNIST Balanced

• Dataset size: 56.63 MiB

• Auto-cached (documentation): Yes

• Splits:

Split	Examples
'test'	18,800
'train'	112,800

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=47),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

emnist/letters

• Config description: EMNIST Letters

• Dataset size: 44.14 MiB

• Auto-cached (documentation): Yes

• Splits:

Split	Examples
'test'	14,800
'train'	88,800

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=37),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

emnist/digits

• Config description: EMNIST Digits

• Dataset size: 120.32 MiB

• Auto-cached (documentation): Yes

• Splits:

Split	Examples
'test'	40,000
'train'	240,000

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

emnist/mnist

• Config description: EMNIST MNIST

• Dataset size: 30.09 MiB

• Auto-cached (documentation): Yes

• Splits:

Split	Examples
'test'	10,000
'train'	60,000

• Feature structure:

FeaturesDict({
    'image': Image(shape=(28, 28, 1), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
image	Image	(28, 28, 1)	uint8
label	ClassLabel		int64

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):