def test_put_probmap_data_dimorder_zxyc(self):
img_path = os.path.join(
base_path, '../test_data/ilastik/dimensionstest/images/*')
label_path = os.path.join(
base_path,
'../test_data/ilastik/dimensionstest/x15_y10_z2_c4_classes2.ilp')
with tempfile.TemporaryDirectory() as tmpdirname:
c = TiffConnector(img_path, 'some/path', savepath=tmpdirname)
d = Dataset(c)
size = (2, 3, 5) # zxy
batch_size = 1
# shape: 1 batch, 4 channels, 2 z, 3 x, 5 y
p = PredictionBatch(d, batch_size, size)
self.assertEqual((1, 4, 2, 3, 5), p[0].pixels().shape)
self.assertEqual((1, 4, 2, 3, 5), p[1].pixels().shape)
p.set_pixel_dimension_order('bzxyc')
self.assertEqual((1, 2, 3, 5, 4), p[0].pixels().shape)
self.assertEqual((1, 2, 3, 5, 4), p[1].pixels().shape)
pixels = p[0].pixels()
for tile in p:
tile.put_probmap_data(pixels)
def test_prediction_loop(self):
# mock classification function
def classify(pixels, value):
return np.ones(pixels.shape) * value
# define data locations
pixel_image_dir = os.path.join(
base_path, '../test_data/tiffconnector_1/im/*.tif')
label_image_dir = os.path.join(
base_path, '../test_data/tiffconnector_1/labels/*.tif')
tile_size = (1, 5, 4) # size of network output layer in zxy
padding = (0, 0, 0) # padding of network input layer in zxy,
# in respect to output layer
# Make training_batch mb and prediction interface p with
# TiffConnector binding.
c = TiffConnector(pixel_image_dir,
label_image_dir,
savepath=self.tmpdir)
p = PredictionBatch(Dataset(c), 2, tile_size, padding_zxy=padding)
self.assertEqual(len(p), 255)
self.assertEqual(p.labels, {1, 2, 3})
# classify the whole bound dataset
for counter, item in enumerate(p):
pixels = item.pixels() # input for classifier
mock_classifier_result = classify(pixels, counter)
# pass classifier results for each class to data source
item.put_probmap_data(mock_classifier_result)
def test_put_probmap_data_when_no_labels_available(self):
img_path = os.path.abspath(
os.path.join(base_path, '../test_data/tiffconnector_1/im/*'))
c = io_connector(img_path, '', savepath=self.tmpdir)
d = Dataset(c)
size = (1, 3, 4)
batch_size = 2
p = PredictionBatch(d, batch_size, size)
data = np.ones((2, 2, 1, 3, 4))
p[0].put_probmap_data(data)
data = np.ones((2, 2, 1, 3, 4))
p[1].put_probmap_data(data)
data = np.ones((2, 2, 1, 3, 4))
p[2].put_probmap_data(data)
val = [
'40width26height3slices_rgb_class_1.tif',
'40width26height3slices_rgb_class_2.tif'
]
self.assertEqual(sorted(os.listdir(self.tmpdir)), val)
def test_getitem_2(self):
img_path = os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
label_path = os.path.join(base_path, '/path/to/nowhere')
c = TiffConnector(img_path, label_path)
d = Dataset(c)
size = (1, 6, 4)
batch_size = 3
p = PredictionBatch(d, batch_size, size)
# batch size is 3, this means all3 tempplates fit in one batch
self.assertEqual(len(p), 1)
self.assertEqual(p[0].pixels().shape, (3, 3, 1, 6, 4))
def test_put_probmap_data_2(self):
img_path = os.path.abspath(
os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
)
label_path = os.path.join(base_path, '/path/to/nowhere')
c = TiffConnector(img_path, label_path, savepath=self.tmpdir)
d = Dataset(c)
size = (1, 2, 2)
batch_size = 1
p = PredictionBatch(d, batch_size, size)
pixel_val = 0
for mb in p:
pixel_val += 10
data = np.ones((1, 2, 1, 2, 2)) * pixel_val
mb.put_probmap_data(data)
pixelmap = memmap(
os.path.join(self.tmpdir, '6width4height3slices_rgb_class_1.tif'))
# zslice 0
val_0 = np.array([[10., 10., 30., 30., 50., 50.],
[10., 10., 30., 30., 50., 50.],
[20., 20., 40., 40., 60., 60.],
[20., 20., 40., 40., 60., 60.]])
assert_array_almost_equal(pixelmap[0, :, :, 0], val_0)
# zslice 1
val_1 = np.array([[70., 70., 90., 90., 110., 110.],
[70., 70., 90., 90., 110., 110.],
[80., 80., 100., 100., 120., 120.],
[80., 80., 100., 100., 120., 120.]])
assert_array_almost_equal(pixelmap[1, :, :, 0], val_1)
# zslice 2
val_2 = np.array([[130., 130., 150., 150., 170., 170.],
[130., 130., 150., 150., 170., 170.],
[140., 140., 160., 160., 180., 180.],
[140., 140., 160., 160., 180., 180.]])
assert_array_almost_equal(pixelmap[2, :, :, 0], val_2)
def test_computepos_3(self):
img_path = os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
label_path = os.path.join(base_path, '/path/to/nowhere')
c = TiffConnector(img_path, label_path)
d = Dataset(c)
size = (2, 6, 4)
batch_size = 1
p = PredictionBatch(d, batch_size, size)
val = [(0, (0, 0, 0)), (0, (1, 0, 0))]
for pos, valpos in zip(p._all_tile_positions, val):
assert_array_equal(pos[1], np.array(valpos[1]))
self.assertEqual(pos[0], valpos[0])
def test_computepos_1(self):
img_path = os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
label_path = os.path.join(base_path, '/path/to/nowhere')
c = TiffConnector(img_path, label_path)
d = Dataset(c)
size = (1, 1, 1)
batch_size = 1
p = PredictionBatch(d, batch_size, size)
self.assertEqual(len(p._all_tile_positions), 6 * 4 * 3)
tilepos = [(p[0], tuple(p[1])) for p in p._all_tile_positions]
self.assertEqual(len(tilepos), len(set(tilepos)))
def test_pixel_dimensions(self):
img_path = os.path.abspath(
os.path.join(base_path, '../test_data/tiffconnector_1/im/*'))
c = io_connector(img_path, '', savepath=self.tmpdir)
d = Dataset(c)
size = (1, 5, 4)
batch_size = 2
p = PredictionBatch(d, batch_size, size)[0]
print(p.pixels().shape)
self.assertEqual((2, 3, 1, 5, 4), p.pixels().shape)
p.set_pixel_dimension_order('bzxyc')
self.assertEqual((2, 1, 5, 4, 3), p.pixels().shape)
def test_put_probmap_data(self):
img_path = os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
label_path = os.path.join(base_path, '/path/to/nowhere')
savepath = tempfile.TemporaryDirectory()
c = TiffConnector(img_path, label_path, savepath=savepath.name)
d = Dataset(c)
size = (1, 6, 4)
batch_size = 1
p = PredictionBatch(d, batch_size, size)
data = np.ones((1, 2, 1, 6, 4))
p[0].put_probmap_data(data)
p[1].put_probmap_data(data)
p[2].put_probmap_data(data)
def test_put_probmap_data_multichannel_label(self):
img_path = os.path.join(base_path, '../test_data/tiffconnector_1/im/*')
label_path = os.path.join(
base_path, '../test_data/tiffconnector_1/labels_multichannel/*')
savepath = tempfile.TemporaryDirectory()
c = TiffConnector(img_path, label_path, savepath=savepath.name)
d = Dataset(c)
original_labels = c.original_label_values_for_all_images()
res = c.calc_label_values_mapping(original_labels)
d = Dataset(c)
size = (1, 3, 4)
batch_size = 1
p = PredictionBatch(d, batch_size, size)
data = np.ones((1, 6, 1, 3, 4))
p[0].put_probmap_data(data)
def test_put_probmap_data_3(self):
img_path = os.path.abspath(
os.path.join(base_path, '../test_data/tiffconnector_1/im/*'))
label_path = os.path.abspath(
os.path.join(base_path, '../test_data/tiffconnector_1/labels/*'))
c = TiffConnector(img_path, label_path, savepath=self.tmpdir)
d = Dataset(c)
size = (1, 3, 4)
batch_size = 2
p = PredictionBatch(d, batch_size, size)
data = np.ones((2, 3, 1, 3, 4))
p[0].put_probmap_data(data)
data = np.ones((2, 3, 1, 3, 4))
p[1].put_probmap_data(data)
data = np.ones((2, 3, 1, 3, 4))
p[2].put_probmap_data(data)
def test_getitem_1(self):
img_path = os.path.join(
base_path,
'../test_data/tiffconnector_1/im/6width4height3slices_rgb.tif')
label_path = os.path.join(base_path, '/path/to/nowhere')
c = TiffConnector(img_path, label_path)
d = Dataset(c)
size = (1, 6, 4)
batch_size = 2
p = PredictionBatch(d, batch_size, size)
# batch size is 2, so the first 2 tiles go with the first batch
# (size two), the third tile in in the second batch. the second
# batch has only size 1 (is smaller than the specified batch size),
# because it contains the rest.
self.assertEqual(len(p), 2)
self.assertEqual(p[0].pixels().shape, (2, 3, 1, 6, 4))
self.assertEqual(p[1].pixels().shape, (1, 3, 1, 6, 4))
def test_put_probmap_data_3(self):
img_path = os.path.join(base_path, '../test_data/tiffconnector_1/im/*')
label_path = os.path.join(base_path,
'../test_data/tiffconnector_1/labels/*')
savepath = tempfile.TemporaryDirectory()
c = TiffConnector(img_path, label_path, savepath=savepath.name)
d = Dataset(c)
size = (1, 3, 4)
batch_size = 2
p = PredictionBatch(d, batch_size, size)
data = np.ones((2, 3, 1, 3, 4))
p[0].put_probmap_data(data)
data = np.ones((2, 3, 1, 3, 4))
p[1].put_probmap_data(data)
data = np.ones((2, 3, 1, 3, 4))
p[2].put_probmap_data(data)