niftynet.layer.rand_flip module

class RandomFlipLayer(flip_axes, flip_probability=0.5, name='random_flip')[source]

Bases: niftynet.layer.base_layer.RandomisedLayer

Add a random flipping layer as pre-processing.

__init__(flip_axes, flip_probability=0.5, name='random_flip')[source]
Parameters:
  • flip_axes – a list of indices over which to flip
  • flip_probability – the probability of performing the flip (default = 0.5)
  • name –
randomise(spatial_rank=3)[source]
layer_op(inputs, interp_orders=None, *args, **kwargs)[source]