def test_chain(self):
class OpA(graph.Operator):
Input = graph.InputSlot() # required slot
def setupOutputs(self):
pass
def propagateDirty(self, *a, **kw):
pass
class OpB(graph.Operator):
Input = graph.InputSlot() # required slot
Output = graph.OutputSlot()
setupOutputs = mock.Mock()
def propagateDirty(self, *a, **kw):
pass
g = graph.Graph()
op_a = OpA(graph=g)
op_b = OpB(graph=g)
op_b.Input.connect(op_a.Input)
with op_a.transaction:
op_a.Input.setValue("fadf")
op_b.setupOutputs.assert_not_called()
op_b.setupOutputs.assert_called_once()
def aaaNumpyFile(self):
g =graph.Graph()
npfile = "/home/akreshuk/data/synapse_small_4d.npy"
reader = OpInputDataReader(graph=g)
reader.FilePath.setValue(npfile)
#out = reader.Output[:].wait()
#print out.shape
opFeatures = OpPixelFeaturesPresmoothed(graph=g)
opFeatures.Scales.setValue(self.scales)
opFeatures.FeatureIds.setValue(self.featureIds)
opFeatures.Input.connect(reader.Output)
opFeatures.Matrix.setValue(self.selectedFeatures[5])
out = opFeatures.Output[:].wait()
print out.shape
opFeaturesInterp = OpPixelFeaturesInterpPresmoothed(graph=g)
opFeaturesInterp.Scales.setValue(self.scales)
opFeaturesInterp.FeatureIds.setValue(self.featureIds)
opFeaturesInterp.Input.connect(reader.Output)
opFeaturesInterp.Matrix.setValue(self.selectedFeatures[5])
opFeaturesInterp.InterpolationScaleZ.setValue(2)
out = opFeaturesInterp.Output[:].wait()
print out.shape
def testSimpleCleanup(self):
g = graph.Graph()
op = OpSimple(graph=g)
r = weakref.ref(op)
del op
gc.collect()
assert r() is None, "cleanup failed"
def testTransactionMultipleSetsOnSameSlot(self):
g = graph.Graph()
op = TransactionOp(graph=g)
with op.transaction:
op.Input1.setValue("val1")
op.Input1.setValue("val2")
op.setupOutputs.assert_called_once()
def setup_method(self, method):
self.g = graph.Graph()
self.op1 = OpWithMultiInputs(graph=self.g)
self.op2 = OpWithMultiInputs(graph=self.g)
self.wrappedOp = OperatorWrapper(OpA, graph=self.g)
self.wrappedOp.Input1.connect(self.op1.Input)
self.wrappedOp.Input2.connect(self.op2.Input)
def testNestedTransactionFails(self):
g = graph.Graph()
op = TransactionOp(graph=g)
with op.transaction:
op.Input1.setValue("val1")
with pytest.raises(AssertionError):
with op.transaction:
op.Input2.setValue("val2")
def testConnectedCleanup(self):
g = graph.Graph()
op1 = OpSimple(graph=g)
op2 = OpSimple(graph=g)
op2.Input.connect(op1.Output)
op2.Input.disconnect()
# op2.cleanUp()
r = weakref.ref(op2)
del op2
gc.collect()
assert r() is None, "cleanup failed"
def test_operator_str():
g = graph.Graph()
class OpA(graph.Operator):
Input = graph.InputSlot(level=2)
op = OpA(graph=g)
op.Input.resize(2)
assert "level=1" in str(op.Input[0])
assert "index=(0,)" in str(op.Input[0])
assert "len=2" in str(op.Input)
assert "index" not in str(op.Input)
def test_operator_except_formatting(self):
op = self.BrokenOp(graph=graph.Graph())
exc = None
try:
op.Out.value
except Exception as e:
exc = e
assert exc
stack = operator.format_operator_stack(exc.__traceback__)
assert stack
assert len(stack) == 2
assert "TestOperatorStackFormatter.BrokenOp.execute" in stack[0]
def aaaAssert(self):
g = graph.Graph()
data = numpy.zeros((self.nx, self.ny, self.nz), dtype=numpy.float32)
for i in range(self.data3d.shape[2]):
data[:, :, i]=i
data = data.view(vigra.VigraArray)
data.axistags = vigra.VigraArray.defaultAxistags(3)
opFeaturesInterp = OpPixelFeaturesInterpPresmoothed(graph=g)
opFeaturesInterp.Input.setValue(data)
opFeaturesInterp.Scales.setValue(self.scales)
opFeaturesInterp.FeatureIds.setValue(self.featureIds)
opFeaturesInterp.InterpolationScaleZ.setValue(self.scaleZ)
opFeaturesInterp.Matrix.setValue(self.selectedFeatures[0])
out = opFeaturesInterp.Output[:].wait()
print "passed"
def runFeatures(self, data, dataInterp):
g = graph.Graph()
opFeatures = OpPixelFeaturesPresmoothed(graph=g)
opFeaturesInterp = OpPixelFeaturesInterpPresmoothed(graph=g)
opFeatures.Input.setValue(dataInterp)
opFeaturesInterp.Input.setValue(data)
opFeatures.Scales.setValue(self.scales)
opFeaturesInterp.Scales.setValue(self.scales)
opFeatures.FeatureIds.setValue(self.featureIds)
opFeaturesInterp.FeatureIds.setValue(self.featureIds)
opFeaturesInterp.InterpolationScaleZ.setValue(self.scaleZ)
#for i, imatrix in enumerate(self.selectedFeatures[0:1]):
for i, imatrix in enumerate(self.selectedFeatures):
opFeatures.Matrix.setValue(imatrix)
opFeaturesInterp.Matrix.setValue(imatrix)
outputInterpData = opFeatures.Output[:].wait()
outputInterpFeatures = opFeaturesInterp.Output[:].wait()
for iz in range(self.nz):
#for iz in range(2, 3):
#print iz, iz*self.scaleZ
try:
outputInterpDataSlice = opFeatures.Output[:, :, iz*self.scaleZ:iz*self.scaleZ+1, :].wait()
outputInterpFeaturesSlice = opFeaturesInterp.Output[:, :, iz, :].wait()
assert_array_almost_equal(outputInterpDataSlice, outputInterpFeaturesSlice, 1)
assert_array_almost_equal(outputInterpData[:, :, iz*self.scaleZ, 0], outputInterpFeatures[:, :, iz, 0], 1)
#assert_array_almost_equal(outputInterpDataSlice[:, :, 0, :], outputInterpData[:, :, iz*self.scaleZ, :], 3)
assert_array_almost_equal(outputInterpFeatures[:, :, iz, :], outputInterpFeaturesSlice[:, :, 0, :], 1)
except AssertionError:
print "failed for feature:", imatrix, i
print "failed for slice:", iz
print "inter data:", outputInterpData[:, :, iz*self.scaleZ, 0]
print "inter features:", outputInterpFeatures[:, :, iz, 0]
print "inter data slice:", outputInterpDataSlice[:, :, 0, 0]
print "inter features:", outputInterpFeaturesSlice[:, :, 0, 0]
raise AssertionError
def aaaSlices(self):
g = graph.Graph()
opFeatures = OpPixelFeaturesPresmoothed(graph=g)
opFeatures.Scales.setValue(self.scales)
opFeatures.FeatureIds.setValue(self.featureIds)
opFeatures.Input.setValue(self.dataNoChannels)
for i, imatrix in enumerate(self.selectedFeatures):
opFeatures.Matrix.setValue(imatrix)
#compute in one piece
dataOne = opFeatures.Output[:].wait()
#compute slice-wise
for z in range(self.nz):
dataSlice = opFeatures.Output[:, :, z:z+1].wait()
try:
assert_array_almost_equal(dataOne[:, :, z:z+1], dataSlice, 2)
except AssertionError:
print "wrong for matrix:", imatrix
print "wrong for slice:", z
print dataOne[:, :, z:z+1]
print dataSlice
raise AssertionError
def testTransactionSetMultipleSlots(self):
input1, input2 = None, None
def fetch_values(self, *args, **kwargs):
nonlocal input1, input2
input1 = self.Input1.value
input2 = self.Input2.value
g = graph.Graph()
op = TransactionOp(graph=g)
setup_mock = mock.Mock()
setup_mock.side_effect = fetch_values
op.setupOutputs = types.MethodType(setup_mock, op)
with op.transaction:
op.Input1.setValue("val1")
op.Input2.setValue("val2")
op.setupOutputs.assert_called_once()
assert input1 == "val1"
assert input2 == "val2"
def __init__(
self,
current_opFeatureSelection,
current_pixelClassificationApplet,
labels_list_data
):
'''
:param current_opFeatureSelection: opFeatureSelection from ilastik
:param current_opPixelClassification: opPixelClassification form Ilastik
'''
super(FeatureSelectionDialog, self).__init__()
self.pixelClassificationApplet = current_pixelClassificationApplet
self.opPixelClassification = current_pixelClassificationApplet.topLevelOperatorView
self.opFeatureSelection = current_opFeatureSelection
self._init_feature_matrix = False
# lazyflow required operator to be connected to a graph, although no interconnection takes place here
g = graph.Graph()
# instantiate feature selection operators
# these operators are not connected to the ilastik lazyflow architecture.
# Reason provided in self._run_selection()
self.opFilterFeatureSelection = opPixelClassification.OpFilterFeatureSelection(graph=g)
self.opWrapperFeatureSelection = opPixelClassification.OpWrapperFeatureSelection(graph=g)
self.opGiniFeatureSelection = opPixelClassification.OpGiniFeatureSelection(graph=g)
# retrieve the featureMatrixCaches operator from the opPixelClassification. This operator provides the features
# and labels matrix required by the feature selection operators
self.opFeatureMatrixCaches = self.opPixelClassification.opFeatureMatrixCaches
'''FIXME / FixMe: the FeatureSelectionDialog will only display one slice of the dataset. This is for RAM saving
reasons. By using only one slice, we can simple predict the segmentation of that slice for each feature set and
store it in RAM. If we allowed to show the whole dataset, then we would have to copy the opFeatureSelection and
opPixelClassification once for each feature set. This would result in too much feature computation time as
well as too much RAM usage.
However, this shortcoming could be overcome by creating something like an opFeatureSubset. Then we would enable
all features in the opFeatureSelection and the feature sets are created by 'filtering' the output of the
opFeatureSelection. Thereby, provided that features in the opFeatureSelection are cached (are they?) the
features would not have to be recalculated for each feature set.'''
self._xysliceID = -1
self._initialized_all_features_segmentation_layer = False
self._initialized_current_features_segmentation_layer = False
self._initialized_feature_matrix = False
self._selected_feature_set_id = None
self.selected_features_matrix = self.opFeatureSelection.SelectionMatrix.value
self.feature_channel_names = None #this gets initialized when the matrix is set to all features in _run_selection
self._stack_dim = self.opPixelClassification.InputImages.meta.shape
self._stack_axistags = self.opPixelClassification.InputImages.meta.axistags
self.__selection_methods = {
0: "gini",
1: "filter",
2: "wrapper"
}
self._selection_params = {
"num_of_feat": 7, #arbitrary number for the default, Ulli thinks it's good
"c": 0.1
}
self._selection_method = "None"
self._gui_initialized = False # is set to true once gui is initialized, prevents multiple initialization
self._feature_selection_results = []
self.labels_list_data = labels_list_data
self.layerstack = layerwidget.LayerStackModel()
# this initializes the actual GUI
self._init_gui()
# set default parameter values
self.number_of_feat_box.setValue(self._selection_params["num_of_feat"])
self.spinbox_c_widget.setValue(self._selection_params["c"])
# connect functionality
self.cancel_button.clicked.connect(self.reject)
self.select_set_button.clicked.connect(self.accept)
self.select_method_cbox.currentIndexChanged.connect(self._handle_selected_method_changed)
self.spinbox_c_widget.valueChanged.connect(self._update_parameters)
self.number_of_feat_box.valueChanged.connect(self._update_parameters)
self.run_button.clicked.connect(self._run_selection)
self.all_feature_sets_combo_box.currentIndexChanged.connect(self._handle_selected_feature_set_changed)
# make sure internal variable are in sync with gui
self._handle_selected_method_changed()
self._update_parameters()
self.resize(1366, 768)
def propagateDirty(self, slot, subindex, roi):
self.Output.setDirty(slice(None))
def cleanup(self):
self.cache.Input.disconnect()
if __name__ == '__main__':
# assumes there is a server running at localhost
logging.basicConfig(level=logging.DEBUG)
volume_url = 'http://localhost:8000/cremi'
from lazyflow import graph
g = graph.Graph()
op = OpRESTfulPrecomputedChunkedVolumeReader(graph=g)
op.BaseUrl.setValue(volume_url)
print(f'available scales: {op.AvailableScales.value}')
print(f'Selected scale: {op.Scale.value}')
# get some data
roi = ((0, 0, 0, 0), (1, 10, 100, 100))
data = op.Output(*roi).wait()
import h5py
with h5py.File('/tmp/temph5.h5', 'w') as f:
f.create_dataset('exported', data=data)
# get some data for the second time, check on server that it has only
# been requested once
data2 = op.Output(*roi).wait()
def setup_method(self, method):
self.g = graph.Graph()
def setup(self):
self.g = graph.Graph()
