dltk package

• dltk.core package
  • dltk.core.activations module
  • dltk.core.losses module
  • dltk.core.metrics module
  • dltk.core.residual_unit module
  • dltk.core.upsample module
  • Module contents
• dltk.io package
  • dltk.io.augmentation module
  • dltk.io.preprocessing module
  • dltk.core.io.abstract_reader module
  • Module contents
• dltk.networks package
  • dltk.networks.autoencoder package
    • dltk.networks.autoencoder.convolutional_autoencoder module
    • Module contents
  • dltk.networks.gan package
    • dltk.networks.gan.dcgan module
    • Module contents
  • dltk.networks.regression_classification package
    • dltk.networks.regression_classification.resnet module
    • Module contents
  • dltk.networks.segmentation package
    • dltk.networks.segmentation.deepmedic module
    • dltk.networks.segmentation.fcn module
    • dltk.networks.segmentation.unet module
    • Module contents
  • dltk.networks.super_resolution package
    • dltk.networks.super_resolution.simple_super_resolution module
    • Module contents
  • Module contents

dltk.utils module

class dltk.utils.SlidingWindow(img_shape, window_shape, has_batch_dim=True, striding=None)

Bases: object

Sliding window iterator which produces slice objects to slice in a sliding window. This is useful for inference.

Constructs a sliding window iterator

Parameters:

• img_shape (array_like) – shape of the image to slide over
• window_shape (array_like) – shape of the window to extract
• has_batch_dim (bool, optional) – flag to indicate whether a batch dimension is present
• striding (array_like, optional) – amount to move the window between each position

next()

dltk.utils.sliding_window_segmentation_inference(session, ops_list, sample_dict, batch_size=1, striding=None)

Utility function to perform sliding window inference for segmentation

Parameters:

• session (tf.Session) – TensorFlow session to run ops with
• ops_list (array_like) – Operators to fetch assemble with sliding window
• sample_dict (dict) – Dictionary with tf.Placeholder keys mapping the
• placeholders to their respective input
• batch_size (int, optional) – Number of sliding windows to batch for calculation
• striding (array_like) – Striding of the sliding window. Can be used to adjust overlap etc.

Returns:

List of np.arrays corresponding to the assembled outputs of

ops_list

Return type:

list

Module contents