datasets

Description

Dataset handles and manipulation methods.

All datasets should be derived from hybrid_learning.datasets.base.BaseDataset which acts as Sequence over tuples of (input, ground_truth).

This module implements a couple of useful

• dataset handles (see custom)

• transformations (see transforms)

• cache handles (see caching)

• visualization helpers (see data_visualization)

Sub-modules

activations_handle	Wrapper dealing with generating and caching activation maps from a dataset.
base	Basic dataset model.
caching	Base classes and implementations of cache handles.
custom	Custom handles for specific datasets.
data_visualization	Helper functions for visualization and analysis of image datasets.
transforms	Dataset modifier (encoder, decoder, transformations) and helper functions.