pydnn.data module

class pydnn.data.DirectoryLabeledImageSet(base_dir, dtype='float32')

Bases: object

Builds training data from a directory where each subdirectory is the name of a class, and contains all the examples of images that class.

Parameters:

• base_dir (string) – the directory containing the class directories
• dtype (string) – the data type to use for the ndarray containing the labels

build(num_images=None)

get_files(num_files=None)

get_labels()

get_random_file(rng)

class pydnn.data.UnlabeledImageSet(base_dir)

Bases: object

Builds an inference set from a directory containing unlabeled images.

Parameters:base_dir (string) – the directory containing the images

build(start, stop)

get_files()