deepscribe.pipeline.selection

class deepscribe.pipeline.selection.SelectDatasetTask(*args, **kwargs)

Selecting classes from larger dataset to produce a smaller archive for training. Assigns data to the train/validation/test sets. Returns .npz file with the different categories.

output()

LocalTarget with path of the output NPZ archive.

Returns:luigi.LocalTarget

requires()

Returns:StandardizeImageSizeTask

run()

Splits data into train, test, eval sets and saves to npz archive.

Returns: