svhn_cropped

• Description:

The Street View House Numbers (SVHN) Dataset is an image digit recognition dataset of over 600,000 digit images coming from real world data. Images are cropped to 32x32.

• Additional Documentation: Explore on Papers With Code

• Homepage: http://ufldl.stanford.edu/housenumbers/

• Source code: tfds.datasets.svhn_cropped.Builder

• Versions:

  • 3.1.0 (default): New split API (https://tensorflow.org/datasets/splits)

• Download size: 1.47 GiB

• Dataset size: 1.09 GiB

• Auto-cached (documentation): No

• Splits:

Split	Examples
'extra'	531,131
'test'	26,032
'train'	73,257

• Feature structure:

FeaturesDict({
    'id': Text(shape=(), dtype=string),
    'image': Image(shape=(32, 32, 3), dtype=uint8),
    'label': ClassLabel(shape=(), dtype=int64, num_classes=10),
})

• Feature documentation:

Feature	Class	Shape	Dtype	Description
	FeaturesDict
id	Text		string
image	Image	(32, 32, 3)	uint8
label	ClassLabel		int64

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

• Figure (tfds.show_examples):

• Examples (tfds.as_dataframe):

• Citation:

"""Street View House Numbers (SVHN) Dataset, cropped version."""

@article{Netzer2011,
author = {Netzer, Yuval and Wang, Tao and Coates, Adam and Bissacco, Alessandro and Wu, Bo and Ng, Andrew Y},
booktitle = {Advances in Neural Information Processing Systems ({NIPS})},
title = {Reading Digits in Natural Images with Unsupervised Feature Learning},
year = {2011}
}