Python API

The MNIST Database is a database of handwritten digits, which has a training set of 60,000 examples, and a test set of 10,000 examples. It is a subset of a larger set available from NIST. The digits have been size-normalized and centered in a fixed-size image. You can download the MNIST database from: http://yann.lecun.com/exdb/mnist/

bob.db.mnist.get_config()

Returns a string containing the configuration information.

class bob.db.mnist.Database(data_dir=None)

Wrapper class for the MNIST database of handwritten digits (http://yann.lecun.com/exdb/mnist/).

Creates the database. The data_dir argument should be the path to the directory containing the four binary files available from http://yann.lecun.com/exdb/mnist/

data(groups=None, labels=None)

Loads the MNIST samples and labels and returns them in NumPy arrays

Keyword Parameters:

groups

One of the groups ‘train’ or ‘test’ or a list with both of them (which is the default).

labels

A subset of the labels (digits 0 to 9) (everything is the default).

Returns: A tuple composed of images and labels as 2D numpy arrays considering

all the filtering criteria and organized as follows:

images

A 2D numpy.ndarray with as many rows as examples in the dataset, as many columns as pixels (actually, there are 28x28 = 784 rows). The pixels of each image are unrolled in C-scan order (i.e., first row 0, then row 1, etc.).

labels

A 1D numpy.ndarray with as many elements as examples in the dataset.

groups()

Returns the vector of groups

labels()

Returns the vector of labels