def test_linear_channels_last(self):
image = np.array(3 * [[
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[1, 1, 1, 1, 1],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0]
]], dtype=DTYPE).transpose(1, 2, 0)
expected = np.array(3 * [[
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[0, 1, 1, 1, 1],
[0, 0, 1, 0, 0],
[0, 0, 0, 0, 0]
]], dtype=DTYPE).transpose(1, 2, 0)
intp = gryds.MultiChannelInterpolator(image, gryds.LinearInterpolator, data_format='channels_last', cval=[0, 0, 0])
trf = gryds.AffineTransformation(ndim=2, angles=[np.pi/2.], center=[0.4, 0.4])
new_image = intp.transform(trf).astype(DTYPE)
np.testing.assert_almost_equal(expected, new_image, decimal=4)
intp = gryds.MultiChannelInterpolator(image, gryds.LinearInterpolator, data_format='channels_last')
trf = gryds.AffineTransformation(ndim=2, angles=[np.pi/2.], center=[0.4, 0.4])
new_image = intp.transform(trf).astype(DTYPE)
np.testing.assert_almost_equal(expected, new_image, decimal=4)
def test_bspline_channels_first(self):
image = np.array(3 * [[
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[1, 1, 1, 1, 1],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0]
]], dtype=DTYPE)
intp = gryds.MultiChannelInterpolator(image, data_format='channels_first', cval=[0, 0, 0])
trf = gryds.AffineTransformation(ndim=2, angles=[np.pi/2.], center=[0.4, 0.4])
new_image = intp.transform(trf, mode='mirror').astype(DTYPE)
np.testing.assert_almost_equal(image, new_image, decimal=4)
intp = gryds.MultiChannelInterpolator(image, data_format='channels_first')
trf = gryds.AffineTransformation(ndim=2, angles=[np.pi/2.], center=[0.4, 0.4])
new_image = intp.transform(trf, mode='mirror').astype(DTYPE)
np.testing.assert_almost_equal(image, new_image, decimal=4)
def test_shape_prop(self):
image = np.array(3 * [[
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0],
[1, 1, 1, 1, 1],
[0, 0, 1, 0, 0],
[0, 0, 1, 0, 0]
]], dtype=DTYPE).transpose(1, 2, 0)
intp = gryds.MultiChannelInterpolator(image, gryds.LinearInterpolator, data_format='channels_last', cval=[0, 0, 0])
self.assertEqual(intp.shape, (5, 5, 3))
def datagenerator(images,
segmentations,
patch_size,
patches_per_im,
batch_size,
augment_brightness=False,
augment_bspline=False):
"""
Simple data-generator to feed patches in batches to the network.
To extract different patches each epoch, steps_per_epoch in fit_generator should be equal to nr_batches.
:param images: Input images
:param segmentations: Corresponding segmentations
:param patch_size: Desired patch size
:param patches_per_im: Amount of patches to extract per image
:param batch_size: Number of patches per batch
:return: Batch of patches to feed to the model
"""
# Total number of patches generated per epoch
total_patches = len(images) * patches_per_im
# Amount of batches in one epoch
nr_batches = int(np.ceil(total_patches / batch_size))
while True:
# Each epoch extract different patches from the training images
x, y = extract_patches(images,
segmentations,
patch_size,
patches_per_im,
seed=np.random.randint(0, 500))
# Feed data in batches to the network
for idx in range(nr_batches):
x_batch = x[idx * batch_size:(idx + 1) * batch_size]
y_batch = y[idx * batch_size:(idx + 1) * batch_size]
## BEGIN EXERCISE
if augment_brightness:
delta = np.random.rand(1)
delta = delta - 0.5
x_batch = x_batch[:] + delta
if augment_bspline:
for jdx in range(x_batch.shape[0]):
b_grid = np.random.rand(2, 3, 3)
b_grid -= 0.5
b_grid /= 10
bspline = gryds.BSplineTransformation(b_grid)
intpl = gryds.MultiChannelInterpolator(x_batch[jdx])
x_batch[jdx] = intpl.transform(bspline)
## END EXERCISE
yield x_batch, y_batch
def gryds_function(images, segmentations):
# input: (images): The input is an array of images
# input: (segmentations): The segmentations of the corresponding input images
# output: (new_image): The output is a geometricly augmented array of images with a randomly defined deformation/ rotation.
# output:(new_segmentation): The output the augmented version of the segmentation
import numpy as np
import gryds
new_image = []
new_segmentation = []
for i in range(len(images)):
affine = gryds.AffineTransformation(
ndim=2,
angles=[
np.pi / 4.
], # List of angles (for 3D transformations you need a list of 3 angles).
center=[0.5, 0.5] # Center of rotation.
)
random_grid = np.random.rand(2, 3, 3)
random_grid -= 0.5
random_grid /= 20
bspline = gryds.BSplineTransformation(random_grid)
# define interpolator of the input_image
interpolator = gryds.MultiChannelInterpolator(images[i],
order=0,
cval=[.1, .2, .3],
mode='nearest')
#define interpolator of the segmentation image
interpolator_segentation = gryds.Interpolator(segmentations[i][:, :,
0],
mode='constant')
#transform the input image
transformed_image = interpolator.transform(bspline, affine)
#transform the segmentation image
transformed_segmentation = interpolator_segentation.transform(
bspline, affine)
#add results into lists
new_segmentation.append(np.clip(transformed_segmentation, 0, 1))
new_image.append(np.clip(transformed_image, 0, 1))
return np.array(new_image), np.array(new_segmentation)
def test_repr(self):
self.assertEqual(
str(gryds.MultiChannelInterpolator(np.random.rand(3, 20, 20))),
'MultiChannelInterpolator(2D, channels_last)')
