def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper( OpFeatureMatrixCache, parent=self )
self._opFeatureMatrixCaches.LabelImage.connect( self.Labels )
self._opFeatureMatrixCaches.FeatureImage.connect( self.Images )
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices( parent=self )
self._opConcatenateFeatureMatrices.FeatureMatrices.connect( self._opFeatureMatrixCaches.LabelAndFeatureMatrix )
self._opConcatenateFeatureMatrices.ProgressSignals.connect( self._opFeatureMatrixCaches.ProgressSignal )
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors( parent=self )
self._opTrainFromFeatures.ClassifierFactory.connect( self.ClassifierFactory )
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect( self._opConcatenateFeatureMatrices.ConcatenatedOutput )
self._opTrainFromFeatures.MaxLabel.connect( self.MaxLabel )
self.Classifier.connect( self._opTrainFromFeatures.Classifier )
# Progress reporting
def _handleFeatureProgress( progress ):
# Note that these progress messages will probably appear out-of-order.
# See comments in OpFeatureMatrixCache
logger.debug("Training: {:02}% (Computing features)".format(int(progress)))
self.progressSignal( 0.8*progress )
self._opConcatenateFeatureMatrices.progressSignal.subscribe( _handleFeatureProgress )
def _handleTrainingComplete():
logger.debug("Training: 100% (Complete)")
self.progressSignal( 100.0 )
self._opTrainFromFeatures.trainingCompleteSignal.subscribe( _handleTrainingComplete )
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper( OpFeatureMatrixCache, parent=self )
self._opFeatureMatrixCaches.LabelImage.connect( self.Labels )
self._opFeatureMatrixCaches.FeatureImage.connect( self.Images )
self._opFeatureMatrixCaches.NonZeroLabelBlocks.connect( self.nonzeroLabelBlocks )
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices( parent=self )
self._opConcatenateFeatureMatrices.FeatureMatrices.connect( self._opFeatureMatrixCaches.LabelAndFeatureMatrix )
self._opConcatenateFeatureMatrices.ProgressSignals.connect( self._opFeatureMatrixCaches.ProgressSignal )
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors( parent=self )
self._opTrainFromFeatures.ClassifierFactory.connect( self.ClassifierFactory )
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect( self._opConcatenateFeatureMatrices.ConcatenatedOutput )
self._opTrainFromFeatures.MaxLabel.connect( self.MaxLabel )
self.Classifier.connect( self._opTrainFromFeatures.Classifier )
# Progress reporting
def _handleFeatureProgress( progress ):
self.progressSignal( 0.8*progress )
self._opConcatenateFeatureMatrices.progressSignal.subscribe( _handleFeatureProgress )
def _handleTrainingComplete():
self.progressSignal( 100.0 )
self._opTrainFromFeatures.trainingCompleteSignal.subscribe( _handleTrainingComplete )
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked,
self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
# Normally, lane removal does not trigger a dirty notification.
# But in this case, if the lane contained any label data whatsoever,
# the classifier needs to be marked dirty.
# We know which slots contain (or contained) label data because they have
# been 'touched' at some point (they became dirty at some point).
self._touched_slots = set()
def handle_new_lane(multislot, index, newlength):
def handle_dirty_lane(slot, roi):
self._touched_slots.add(slot)
multislot[index].notifyDirty(handle_dirty_lane)
self.Labels.notifyInserted(handle_new_lane)
def handle_remove_lane(multislot, index, newlength):
# If the lane we're removing contained
# label data, then mark the downstream dirty
if multislot[index] in self._touched_slots:
self.Classifier.setDirty()
self._touched_slots.remove(multislot[index])
self.Labels.notifyRemove(handle_remove_lane)
class OpTrainVectorwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
MaxLabel = InputSlot()
Classifier = OutputSlot()
# Images[N] --- MaxLabel ------
# \ \
# Labels[N] --> opFeatureMatrixCaches ---(FeatureImage[N])---> opConcatenateFeatureImages ---(label+feature matrix)---> OpTrainFromFeatures ---(Classifier)--->
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper(OpFeatureMatrixCache, parent=self)
self._opFeatureMatrixCaches.LabelImage.connect(self.Labels)
self._opFeatureMatrixCaches.FeatureImage.connect(self.Images)
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices(parent=self)
self._opConcatenateFeatureMatrices.FeatureMatrices.connect(self._opFeatureMatrixCaches.LabelAndFeatureMatrix)
self._opConcatenateFeatureMatrices.ProgressSignals.connect(self._opFeatureMatrixCaches.ProgressSignal)
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors(parent=self)
self._opTrainFromFeatures.ClassifierFactory.connect(self.ClassifierFactory)
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect(self._opConcatenateFeatureMatrices.ConcatenatedOutput)
self._opTrainFromFeatures.MaxLabel.connect(self.MaxLabel)
self.Classifier.connect(self._opTrainFromFeatures.Classifier)
# Progress reporting
def _handleFeatureProgress(progress):
# Note that these progress messages will probably appear out-of-order.
# See comments in OpFeatureMatrixCache
logger.debug("Training: {:02}% (Computing features)".format(int(progress)))
self.progressSignal(0.8 * progress)
self._opConcatenateFeatureMatrices.progressSignal.subscribe(_handleFeatureProgress)
def _handleTrainingComplete():
logger.debug("Training: 100% (Complete)")
self.progressSignal(100.0)
self._opTrainFromFeatures.trainingCompleteSignal.subscribe(_handleTrainingComplete)
def cleanUp(self):
self.progressSignal.clean()
self.Classifier.disconnect()
super(OpTrainVectorwiseClassifierBlocked, self).cleanUp()
def setupOutputs(self):
pass # Nothing to do; our output is connected to an internal operator.
def execute(self, slot, subindex, roi, result):
assert False, "Shouldn't get here..."
def propagateDirty(self, slot, subindex, roi):
pass
class OpTrainVectorwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
MaxLabel = InputSlot()
Classifier = OutputSlot()
# Images[N] --- MaxLabel ------
# \ \
# Labels[N] --> opFeatureMatrixCaches ---(FeatureImage[N])---> opConcatenateFeatureImages ---(label+feature matrix)---> OpTrainFromFeatures ---(Classifier)--->
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper(OpFeatureMatrixCache, parent=self)
self._opFeatureMatrixCaches.LabelImage.connect(self.Labels)
self._opFeatureMatrixCaches.FeatureImage.connect(self.Images)
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices(parent=self)
self._opConcatenateFeatureMatrices.FeatureMatrices.connect(self._opFeatureMatrixCaches.LabelAndFeatureMatrix)
self._opConcatenateFeatureMatrices.ProgressSignals.connect(self._opFeatureMatrixCaches.ProgressSignal)
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors(parent=self)
self._opTrainFromFeatures.ClassifierFactory.connect(self.ClassifierFactory)
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect(self._opConcatenateFeatureMatrices.ConcatenatedOutput)
self._opTrainFromFeatures.MaxLabel.connect(self.MaxLabel)
self.Classifier.connect(self._opTrainFromFeatures.Classifier)
# Progress reporting
def _handleFeatureProgress(progress):
# Note that these progress messages will probably appear out-of-order.
# See comments in OpFeatureMatrixCache
logger.debug("Training: {:02}% (Computing features)".format(int(progress)))
self.progressSignal(0.8 * progress)
self._opConcatenateFeatureMatrices.progressSignal.subscribe(_handleFeatureProgress)
def _handleTrainingComplete():
logger.debug("Training: 100% (Complete)")
self.progressSignal(100.0)
self._opTrainFromFeatures.trainingCompleteSignal.subscribe(_handleTrainingComplete)
def cleanUp(self):
self.progressSignal.clean()
self.Classifier.disconnect()
super(OpTrainVectorwiseClassifierBlocked, self).cleanUp()
def setupOutputs(self):
pass # Nothing to do; our output is connected to an internal operator.
def execute(self, slot, subindex, roi, result):
assert False, "Shouldn't get here..."
def propagateDirty(self, slot, subindex, roi):
pass
class OpTrainVectorwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1) # TODO: Eliminate this slot. It isn't used any more...
MaxLabel = InputSlot()
Classifier = OutputSlot()
# Images[N] --- MaxLabel ------
# \ \
# Labels[N] --> opFeatureMatrixCaches ---(FeatureImage[N])---> opConcatenateFeatureImages ---(label+feature matrix)---> OpTrainFromFeatures ---(Classifier)--->
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper( OpFeatureMatrixCache, parent=self )
self._opFeatureMatrixCaches.LabelImage.connect( self.Labels )
self._opFeatureMatrixCaches.FeatureImage.connect( self.Images )
self._opFeatureMatrixCaches.NonZeroLabelBlocks.connect( self.nonzeroLabelBlocks )
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices( parent=self )
self._opConcatenateFeatureMatrices.FeatureMatrices.connect( self._opFeatureMatrixCaches.LabelAndFeatureMatrix )
self._opConcatenateFeatureMatrices.ProgressSignals.connect( self._opFeatureMatrixCaches.ProgressSignal )
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors( parent=self )
self._opTrainFromFeatures.ClassifierFactory.connect( self.ClassifierFactory )
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect( self._opConcatenateFeatureMatrices.ConcatenatedOutput )
self._opTrainFromFeatures.MaxLabel.connect( self.MaxLabel )
self.Classifier.connect( self._opTrainFromFeatures.Classifier )
# Progress reporting
def _handleFeatureProgress( progress ):
self.progressSignal( 0.8*progress )
self._opConcatenateFeatureMatrices.progressSignal.subscribe( _handleFeatureProgress )
def _handleTrainingComplete():
self.progressSignal( 100.0 )
self._opTrainFromFeatures.trainingCompleteSignal.subscribe( _handleTrainingComplete )
def cleanUp(self):
self.progressSignal.clean()
self.Classifier.disconnect()
super( OpTrainVectorwiseClassifierBlocked, self ).cleanUp()
def setupOutputs(self):
pass # Nothing to do; our output is connected to an internal operator.
def execute(self, slot, subindex, roi, result):
assert False, "Shouldn't get here..."
def propagateDirty(self, slot, subindex, roi):
pass
class OpTrainVectorwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1) # TODO: Eliminate this slot. It isn't used any more...
MaxLabel = InputSlot()
Classifier = OutputSlot()
# Images[N] --- MaxLabel ------
# \ \
# Labels[N] --> opFeatureMatrixCaches ---(FeatureImage[N])---> opConcatenateFeatureImages ---(label+feature matrix)---> OpTrainFromFeatures ---(Classifier)--->
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper( OpFeatureMatrixCache, parent=self )
self._opFeatureMatrixCaches.LabelImage.connect( self.Labels )
self._opFeatureMatrixCaches.FeatureImage.connect( self.Images )
self._opFeatureMatrixCaches.NonZeroLabelBlocks.connect( self.nonzeroLabelBlocks )
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices( parent=self )
self._opConcatenateFeatureMatrices.FeatureMatrices.connect( self._opFeatureMatrixCaches.LabelAndFeatureMatrix )
self._opConcatenateFeatureMatrices.ProgressSignals.connect( self._opFeatureMatrixCaches.ProgressSignal )
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors( parent=self )
self._opTrainFromFeatures.ClassifierFactory.connect( self.ClassifierFactory )
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect( self._opConcatenateFeatureMatrices.ConcatenatedOutput )
self._opTrainFromFeatures.MaxLabel.connect( self.MaxLabel )
self.Classifier.connect( self._opTrainFromFeatures.Classifier )
# Progress reporting
def _handleFeatureProgress( progress ):
self.progressSignal( 0.8*progress )
self._opConcatenateFeatureMatrices.progressSignal.subscribe( _handleFeatureProgress )
def _handleTrainingComplete():
self.progressSignal( 100.0 )
self._opTrainFromFeatures.trainingCompleteSignal.subscribe( _handleTrainingComplete )
def cleanUp(self):
self.progressSignal.clean()
self.Classifier.disconnect()
super( OpTrainVectorwiseClassifierBlocked, self ).cleanUp()
def setupOutputs(self):
pass # Nothing to do; our output is connected to an internal operator.
def execute(self, slot, subindex, roi, result):
assert False, "Shouldn't get here..."
def propagateDirty(self, slot, subindex, roi):
pass
def __init__(self, *args, **kwargs):
super(OpTrainCounter, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._svr = SVR()
params = self._svr.get_params()
self.initInputs(params)
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (self.numRegressors, )
# Normally, lane removal does not trigger a dirty notification.
# But in this case, if the lane contained any label data whatsoever,
# the classifier needs to be marked dirty.
# We know which slots contain (or contained) label data because they have
# been 'touched' at some point (they became dirty at some point).
self._touched_slots = set()
def handle_new_lane(multislot, index, newlength):
def handle_dirty_lane(slot, roi):
self._touched_slots.add(slot)
multislot[index].notifyDirty(handle_dirty_lane)
self.ForegroundLabels.notifyInserted(handle_new_lane)
self.BackgroundLabels.notifyInserted(handle_new_lane)
def handle_remove_lane(multislot, index, newlength):
# If the lane we're removing contained
# label data, then mark the downstream dirty
if multislot[index] in self._touched_slots:
self.Classifier.setDirty()
self._touched_slots.remove(multislot[index])
self.ForegroundLabels.notifyRemove(handle_remove_lane)
self.BackgroundLabels.notifyRemove(handle_remove_lane)
def __init__(self, *args, **kwargs):
super(OpTrainVectorwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._opFeatureMatrixCaches = OperatorWrapper( OpFeatureMatrixCache, parent=self )
self._opFeatureMatrixCaches.LabelImage.connect( self.Labels )
self._opFeatureMatrixCaches.FeatureImage.connect( self.Images )
self._opFeatureMatrixCaches.NonZeroLabelBlocks.connect( self.nonzeroLabelBlocks )
self._opConcatenateFeatureMatrices = OpConcatenateFeatureMatrices( parent=self )
self._opConcatenateFeatureMatrices.FeatureMatrices.connect( self._opFeatureMatrixCaches.LabelAndFeatureMatrix )
self._opConcatenateFeatureMatrices.ProgressSignals.connect( self._opFeatureMatrixCaches.ProgressSignal )
self._opTrainFromFeatures = OpTrainClassifierFromFeatureVectors( parent=self )
self._opTrainFromFeatures.ClassifierFactory.connect( self.ClassifierFactory )
self._opTrainFromFeatures.LabelAndFeatureMatrix.connect( self._opConcatenateFeatureMatrices.ConcatenatedOutput )
self._opTrainFromFeatures.MaxLabel.connect( self.MaxLabel )
self.Classifier.connect( self._opTrainFromFeatures.Classifier )
# Progress reporting
def _handleFeatureProgress( progress ):
self.progressSignal( 0.8*progress )
self._opConcatenateFeatureMatrices.progressSignal.subscribe( _handleFeatureProgress )
def _handleTrainingComplete():
self.progressSignal( 100.0 )
self._opTrainFromFeatures.trainingCompleteSignal.subscribe( _handleTrainingComplete )
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
# Normally, lane removal does not trigger a dirty notification.
# But in this case, if the lane contained any label data whatsoever,
# the classifier needs to be marked dirty.
# We know which slots contain (or contained) label data because they have
# been 'touched' at some point (they became dirty at some point).
self._touched_slots = set()
def handle_new_lane(multislot, index, newlength):
def handle_dirty_lane(slot, roi):
self._touched_slots.add(slot)
multislot[index].notifyDirty(handle_dirty_lane)
self.Labels.notifyInserted(handle_new_lane)
def handle_remove_lane(multislot, index, newlength):
# If the lane we're removing contained
# label data, then mark the downstream dirty
if multislot[index] in self._touched_slots:
self.Classifier.setDirty()
self._touched_slots.remove(multislot[index])
self.Labels.notifyRemove(handle_remove_lane)
def __init__(self, *args, **kwargs):
super(OpTrainClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._mode = None
# Fully connect the vectorwise training operator
self._opVectorwiseTrain = OpTrainVectorwiseClassifierBlocked(
parent=self)
self._opVectorwiseTrain.Images.connect(self.Images)
self._opVectorwiseTrain.Labels.connect(self.Labels)
self._opVectorwiseTrain.ClassifierFactory.connect(
self.ClassifierFactory)
self._opVectorwiseTrain.nonzeroLabelBlocks.connect(
self.nonzeroLabelBlocks)
self._opVectorwiseTrain.MaxLabel.connect(self.MaxLabel)
self._opVectorwiseTrain.progressSignal.subscribe(self.progressSignal)
# Fully connect the pixelwise training operator
self._opPixelwiseTrain = OpTrainPixelwiseClassifierBlocked(parent=self)
self._opPixelwiseTrain.Images.connect(self.Images)
self._opPixelwiseTrain.Labels.connect(self.Labels)
self._opPixelwiseTrain.ClassifierFactory.connect(
self.ClassifierFactory)
self._opPixelwiseTrain.nonzeroLabelBlocks.connect(
self.nonzeroLabelBlocks)
self._opPixelwiseTrain.MaxLabel.connect(self.MaxLabel)
self._opPixelwiseTrain.progressSignal.subscribe(self.progressSignal)
def __init__(self, *args, **kwargs):
super(OpBatchIoSelective, self).__init__(*args, **kwargs)
self.Dirty.meta.shape = (1, )
self.Dirty.meta.dtype = bool
self.OutputDataPath.meta.shape = (1, )
self.OutputDataPath.meta.dtype = object
self.ExportResult.meta.shape = (1, )
self.ExportResult.meta.dtype = object
# Provide default values
self.ExportDirectory.setValue('')
self.Format.setValue(ExportFormat.H5)
self.Suffix.setValue('_results')
self.Dirty.setValue(True)
self.progressSignal = OrderedSignal()
self.ProgressSignal.setValue(self.progressSignal)
self._createDirLock = threading.Lock()
#make a cache of the input image not to request too much
self.ImageCache = OpBlockedArrayCache(parent=self)
self.ImageCache.fixAtCurrent.setValue(False)
self.ImageCache.Input.connect(self.ImageToExport)
def __init__(self, *args, **kwargs):
super(OpTrainCounter, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._svr = SVR()
params = self._svr.get_params()
self.initInputs(params)
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (self.numRegressors,)
def __init__(
self,
h5N5File=None,
h5N5Path=None,
Image=None,
BatchSize: int = None,
CompressionEnabled: bool = None,
*args,
**kwargs,
):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
self.h5N5File.setOrConnectIfAvailable(h5N5File)
self.h5N5Path.setOrConnectIfAvailable(h5N5Path)
self.Image.setOrConnectIfAvailable(Image)
self.BatchSize.setOrConnectIfAvailable(BatchSize)
self.CompressionEnabled.setOrConnectIfAvailable(CompressionEnabled)
def __init__(self, *args, **kwargs):
super(OpTrainSupervoxelClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._mode = None
# Fully connect the vectorwise training operator
self._opVectorwiseTrain = OpTrainSupervoxelwiseClassifierBlocked(parent=self)
self._opVectorwiseTrain.Images.connect(self.Images)
self._opVectorwiseTrain.SupervoxelSegmentation.connect(self.SupervoxelSegmentation)
self._opVectorwiseTrain.SupervoxelFeatures.connect(self.SupervoxelFeatures)
self._opVectorwiseTrain.SupervoxelLabels.connect(self.SupervoxelLabels)
self._opVectorwiseTrain.Labels.connect(self.Labels)
self._opVectorwiseTrain.ClassifierFactory.connect(self.ClassifierFactory)
self._opVectorwiseTrain.progressSignal.subscribe(self.progressSignal)
def __init__(self, *args, **kwargs):
super(OpBatchIo, self).__init__(*args, **kwargs)
self._opExportedImageProvider = None
self.Dirty.meta.shape = (1, )
self.Dirty.meta.dtype = bool
self.OutputDataPath.meta.shape = (1, )
self.OutputDataPath.meta.dtype = object
self.ExportResult.meta.shape = (1, )
self.ExportResult.meta.dtype = object
# Default to Dirty
self.Dirty.setValue(True)
self.progressSignal = OrderedSignal()
self.ProgressSignal.setValue(self.progressSignal)
self._createDirLock = threading.Lock()
def __init__(self, *args, **kwargs):
super(OpBatchIo, self).__init__(*args, **kwargs)
self.Dirty.meta.shape = (1, )
self.Dirty.meta.dtype = bool
self.OutputDataPath.meta.shape = (1, )
self.OutputDataPath.meta.dtype = object
self.ExportResult.meta.shape = (1, )
self.ExportResult.meta.dtype = object
# Provide default values
self.ExportDirectory.setValue('')
self.Format.setValue(ExportFormat.H5)
self.Suffix.setValue('_results')
self.Dirty.setValue(True)
self.progressSignal = OrderedSignal()
self.ProgressSignal.setValue(self.progressSignal)
self._createDirLock = threading.Lock()
class OpH5WriterBigDataset(Operator):
name = "H5 File Writer BigDataset"
category = "Output"
inputSlots = [InputSlot("hdf5File"), # Must be an already-open hdf5File (or group) for writing to
InputSlot("hdf5Path", stype = "string"),
InputSlot("Image"),
InputSlot("CompressionEnabled", value=True)]
outputSlots = [OutputSlot("WriteImage")]
loggingName = __name__ + ".OpH5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
def cleanUp(self):
super( OpH5WriterBigDataset, self ).cleanUp()
# Discard the reference to the dataset, to ensure that hdf5 can close the file.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1,)
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["hdf5File"].value
hdf5Path = self.inputs["hdf5Path"].value
# On windows, there may be backslashes.
hdf5Path = hdf5Path.replace('\\', '/')
hdf5GroupName, datasetName = os.path.split(hdf5Path)
if hdf5GroupName == "":
g = self.f
else:
if hdf5GroupName in self.f:
g = self.f[hdf5GroupName]
else:
g = self.f.create_group(hdf5GroupName)
dataShape=self.Image.meta.shape
taggedShape = self.Image.meta.getTaggedShape()
dtype = self.Image.meta.dtype
if type(dtype) is numpy.dtype:
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
numChannels = 1
if 'c' in taggedShape:
numChannels = taggedShape['c']
# Set up our chunk shape: Aim for a cube that's roughly 300k in size
dtypeBytes = dtype().nbytes
cubeDim = math.pow( 300000 / (numChannels * dtypeBytes), (1/3.0) )
cubeDim = int(cubeDim)
chunkDims = {}
chunkDims['t'] = 1
chunkDims['x'] = cubeDim
chunkDims['y'] = cubeDim
chunkDims['z'] = cubeDim
chunkDims['c'] = numChannels
# h5py guide to chunking says chunks of 300k or less "work best"
assert chunkDims['x'] * chunkDims['y'] * chunkDims['z'] * numChannels * dtypeBytes <= 300000
chunkShape = ()
for i in range( len(dataShape) ):
axisKey = self.Image.meta.axistags[i].key
# Chunk shape can't be larger than the data shape
chunkShape += ( min( chunkDims[axisKey], dataShape[i] ), )
self.chunkShape = chunkShape
if datasetName in g.keys():
del g[datasetName]
kwargs = { 'shape' : dataShape, 'dtype' : dtype, 'chunks' : self.chunkShape }
if self.CompressionEnabled.value:
kwargs['compression'] = 'gzip' # <-- Would be nice to use lzf compression here, but that is h5py-specific.
kwargs['compression_opts'] = 1 # <-- Optimize for speed, not disk space.
self.d=g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs['drange'] = self.Image.meta.drange
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
slicings=self.computeRequestSlicings()
numSlicings = len(slicings)
self.logger.debug( "Dividing work into {} pieces".format( len(slicings) ) )
# Throttle: Only allow 10 outstanding requests at a time.
# Otherwise, the whole set of requests can be outstanding and use up ridiculous amounts of memory.
activeRequests = deque()
activeSlicings = deque()
# Start by activating 10 requests
for i in range( min(10, len(slicings)) ):
s = slicings.pop()
activeSlicings.append(s)
self.logger.debug( "Creating request for slicing {}".format(s) )
activeRequests.append( self.inputs["Image"][s] )
counter = 0
while len(activeRequests) > 0:
# Wait for a request to finish
req = activeRequests.popleft()
s=activeSlicings.popleft()
data = req.wait()
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=s)
else:
self.d[s] = data
req.clean() # Discard the data in the request and allow its children to be garbage collected.
if len(slicings) > 0:
# Create a new active request
s = slicings.pop()
activeSlicings.append(s)
activeRequests.append( self.inputs["Image"][s] )
# Since requests finish in an arbitrary order (but we always block for them in the same order),
# this progress feedback will not be smooth. It's the best we can do for now.
self.progressSignal( 100*counter/numSlicings )
self.logger.debug( "request {} out of {} executed".format( counter, numSlicings ) )
counter += 1
# Save the axistags as a dataset attribute
self.d.attrs['axistags'] = self.Image.meta.axistags.toJSON()
# Be paranoid: Flush right now.
self.f.file.flush()
# We're finished.
result[0] = True
self.progressSignal(100)
def computeRequestSlicings(self):
#TODO: reimplement the request better
shape=numpy.asarray(self.inputs['Image'].meta.shape)
chunkShape = numpy.asarray(self.chunkShape)
# Choose a request shape that is a multiple of the chunk shape
axistags = self.Image.meta.axistags
multipliers = { 'x':5, 'y':5, 'z':5, 't':1, 'c':100 } # For most problems, there is little advantage to breaking up the channels.
multiplier = [multipliers[tag.key] for tag in axistags ]
shift = chunkShape * numpy.array(multiplier)
shift=numpy.minimum(shift,shape)
start=numpy.asarray([0]*len(shape))
stop=shift
reqList=[]
#shape = shape - (numpy.mod(numpy.asarray(shape),
# shift))
for indices in product(*[range(0, stop, step)
for stop,step in zip(shape, shift)]):
start=numpy.asarray(indices)
stop=numpy.minimum(start+shift,shape)
reqList.append(roiToSlice(start,stop))
return reqList
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def setupOutputs(self):
for slot in list(self.Images) + list(self.Labels):
assert slot.meta.getAxisKeys()[-1] == 'c', \
"This opearator assumes channel is the last axis."
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1,)
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super( OpTrainPixelwiseClassifierBlocked, self ).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert issubclass(type(classifier_factory), LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
block_label_roi = sliceToRoi( block_slicing, image_slot.meta.shape )
# Ask for the halo needed by the classifier
axiskeys = image_slot.meta.getAxisKeys()
halo_shape = classifier_factory.get_halo_shape(axiskeys)
assert len(halo_shape) == len( block_label_roi[0] )
assert halo_shape[-1] == 0, "Didn't expect a non-zero halo for channel dimension."
# Expand block by halo, then clip to image bounds
block_label_roi = numpy.array( block_label_roi )
block_label_roi[0] -= halo_shape
block_label_roi[1] += halo_shape
block_label_roi = getIntersection( block_label_roi, roiFromShape(image_slot.meta.shape) )
block_image_roi = numpy.array( block_label_roi )
assert (block_image_roi[:, -1] == [0,1]).all()
num_channels = image_slot.meta.shape[-1]
block_image_roi[:, -1] = [0, num_channels]
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
block_label_data = numpy.asarray( label_slot(*block_label_roi).wait() )
block_image_data = numpy.asarray( image_slot(*block_image_roi).wait() )
label_data_blocks.append( block_label_data )
image_data_blocks.append( block_image_data )
logger.debug("Training new classifier: {}".format( classifier_factory.description ))
classifier = classifier_factory.create_and_train_pixelwise( image_data_blocks, label_data_blocks )
assert issubclass(type(classifier), LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
result[0] = classifier
return result
def propagateDirty(self, slot, subindex, roi):
self.Classifier.setDirty()
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
# Normally, lane removal does not trigger a dirty notification.
# But in this case, if the lane contained any label data whatsoever,
# the classifier needs to be marked dirty.
# We know which slots contain (or contained) label data because they have
# been 'touched' at some point (they became dirty at some point).
self._touched_slots = set()
def handle_new_lane( multislot, index, newlength ):
def handle_dirty_lane( slot, roi ):
self._touched_slots.add(slot)
multislot[index].notifyDirty( handle_dirty_lane )
self.Labels.notifyInserted( handle_new_lane )
def handle_remove_lane( multislot, index, newlength ):
# If the lane we're removing contained
# label data, then mark the downstream dirty
if multislot[index] in self._touched_slots:
self.Classifier.setDirty()
self._touched_slots.remove(multislot[index])
self.Labels.notifyRemove( handle_remove_lane )
def setupOutputs(self):
for slot in list(self.Images) + list(self.Labels):
assert slot.meta.getAxisKeys()[-1] == 'c', \
"This opearator assumes channel is the last axis."
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1,)
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super( OpTrainPixelwiseClassifierBlocked, self ).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert issubclass(type(classifier_factory), LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
# Get labels
block_label_roi = sliceToRoi( block_slicing, label_slot.meta.shape )
block_label_data = label_slot(*block_label_roi).wait()
# Shrink roi to bounding box of actual label pixels
bb_roi_within_block = nonzero_bounding_box(block_label_data)
block_label_bb_roi = bb_roi_within_block + block_label_roi[0]
# Double-check that there is at least 1 non-zero label in the block.
if (block_label_bb_roi[1] > block_label_bb_roi[0]).all():
# Ask for the halo needed by the classifier
axiskeys = image_slot.meta.getAxisKeys()
halo_shape = classifier_factory.get_halo_shape(axiskeys)
assert len(halo_shape) == len( block_label_roi[0] )
assert halo_shape[-1] == 0, "Didn't expect a non-zero halo for channel dimension."
# Expand block by halo, but keep clipped to image bounds
padded_label_roi, bb_roi_within_padded = enlargeRoiForHalo( *block_label_bb_roi,
shape=label_slot.meta.shape,
sigma=halo_shape,
window=1,
return_result_roi=True )
# Copy labels to new array, which has size == bounding-box + halo
padded_label_data = numpy.zeros( padded_label_roi[1] - padded_label_roi[0], label_slot.meta.dtype )
padded_label_data[roiToSlice(*bb_roi_within_padded)] = block_label_data[roiToSlice(*bb_roi_within_block)]
padded_image_roi = numpy.array( padded_label_roi )
assert (padded_image_roi[:, -1] == [0,1]).all()
num_channels = image_slot.meta.shape[-1]
padded_image_roi[:, -1] = [0, num_channels]
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
padded_image_data = numpy.asarray( image_slot(*padded_image_roi).wait() )
label_data_blocks.append( padded_label_data )
image_data_blocks.append( padded_image_data )
if len(image_data_blocks) == 0:
result[0] = None
else:
channel_names = self.Images[0].meta.channel_names
axistags = self.Images[0].meta.axistags
logger.debug("Training new pixelwise classifier: {}".format( classifier_factory.description ))
classifier = classifier_factory.create_and_train_pixelwise( image_data_blocks, label_data_blocks, axistags, channel_names )
result[0] = classifier
if classifier is not None:
assert issubclass(type(classifier), LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
def propagateDirty(self, slot, subindex, roi):
self.Classifier.setDirty()
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.opStackLoader = OpStackLoader(parent=self)
self.opStackLoader.globstring.connect(self.GlobString)
class OpH5WriterBigDataset(Operator):
name = "H5 File Writer BigDataset"
category = "Output"
inputSlots = [InputSlot("hdf5File"), # Must be an already-open hdf5File (or group) for writing to
InputSlot("hdf5Path", stype = "string"),
InputSlot("Image"),
InputSlot("CompressionEnabled", value=True)]
outputSlots = [OutputSlot("WriteImage")]
loggingName = __name__ + ".OpH5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
def cleanUp(self):
super( OpH5WriterBigDataset, self ).cleanUp()
# Discard the reference to the dataset, to ensure that hdf5 can close the file.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1,)
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["hdf5File"].value
hdf5Path = self.inputs["hdf5Path"].value
# On windows, there may be backslashes.
hdf5Path = hdf5Path.replace('\\', '/')
hdf5GroupName, datasetName = os.path.split(hdf5Path)
if hdf5GroupName == "":
g = self.f
else:
if hdf5GroupName in self.f:
g = self.f[hdf5GroupName]
else:
g = self.f.create_group(hdf5GroupName)
dataShape=self.Image.meta.shape
taggedShape = self.Image.meta.getTaggedShape()
dtype = self.Image.meta.dtype
if type(dtype) is numpy.dtype:
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
numChannels = 1
if 'c' in taggedShape:
numChannels = taggedShape['c']
# Set up our chunk shape: Aim for a cube that's roughly 300k in size
dtypeBytes = dtype().nbytes
cubeDim = math.pow( 300000 / (numChannels * dtypeBytes), (1/3.0) )
cubeDim = int(cubeDim)
chunkDims = {}
chunkDims['t'] = 1
chunkDims['x'] = cubeDim
chunkDims['y'] = cubeDim
chunkDims['z'] = cubeDim
chunkDims['c'] = numChannels
# h5py guide to chunking says chunks of 300k or less "work best"
assert chunkDims['x'] * chunkDims['y'] * chunkDims['z'] * numChannels * dtypeBytes <= 300000
chunkShape = ()
for i in range( len(dataShape) ):
axisKey = self.Image.meta.axistags[i].key
# Chunk shape can't be larger than the data shape
chunkShape += ( min( chunkDims[axisKey], dataShape[i] ), )
self.chunkShape = chunkShape
if datasetName in g.keys():
del g[datasetName]
kwargs = { 'shape' : dataShape, 'dtype' : dtype, 'chunks' : self.chunkShape }
if self.CompressionEnabled.value:
kwargs['compression'] = 'gzip' # <-- Would be nice to use lzf compression here, but that is h5py-specific.
kwargs['compression_opts'] = 1 # <-- Optimize for speed, not disk space.
self.d=g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs['drange'] = self.Image.meta.drange
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
slicings=self.computeRequestSlicings()
numSlicings = len(slicings)
self.logger.debug( "Dividing work into {} pieces".format( len(slicings) ) )
# Throttle: Only allow 10 outstanding requests at a time.
# Otherwise, the whole set of requests can be outstanding and use up ridiculous amounts of memory.
activeRequests = deque()
activeSlicings = deque()
# Start by activating 10 requests
for i in range( min(10, len(slicings)) ):
s = slicings.pop()
activeSlicings.append(s)
self.logger.debug( "Creating request for slicing {}".format(s) )
activeRequests.append( self.inputs["Image"][s] )
counter = 0
while len(activeRequests) > 0:
# Wait for a request to finish
req = activeRequests.popleft()
s=activeSlicings.popleft()
data = req.wait()
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=s)
else:
self.d[s] = data
req.clean() # Discard the data in the request and allow its children to be garbage collected.
if len(slicings) > 0:
# Create a new active request
s = slicings.pop()
activeSlicings.append(s)
activeRequests.append( self.inputs["Image"][s] )
# Since requests finish in an arbitrary order (but we always block for them in the same order),
# this progress feedback will not be smooth. It's the best we can do for now.
self.progressSignal( 100*counter/numSlicings )
self.logger.debug( "request {} out of {} executed".format( counter, numSlicings ) )
counter += 1
# Save the axistags as a dataset attribute
self.d.attrs['axistags'] = self.Image.meta.axistags.toJSON()
# Be paranoid: Flush right now.
self.f.file.flush()
# We're finished.
result[0] = True
self.progressSignal(100)
def computeRequestSlicings(self):
#TODO: reimplement the request better
shape=numpy.asarray(self.inputs['Image'].meta.shape)
chunkShape = numpy.asarray(self.chunkShape)
# Choose a request shape that is a multiple of the chunk shape
axistags = self.Image.meta.axistags
multipliers = { 'x':5, 'y':5, 'z':5, 't':1, 'c':100 } # For most problems, there is little advantage to breaking up the channels.
multiplier = [multipliers[tag.key] for tag in axistags ]
shift = chunkShape * numpy.array(multiplier)
shift=numpy.minimum(shift,shape)
start=numpy.asarray([0]*len(shape))
stop=shift
reqList=[]
#shape = shape - (numpy.mod(numpy.asarray(shape),
# shift))
for indices in product(*[range(0, stop, step)
for stop,step in zip(shape, shift)]):
start=numpy.asarray(indices)
stop=numpy.minimum(start+shift,shape)
reqList.append(roiToSlice(start,stop))
return reqList
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def __init__(self, *args, **kwargs):
super(OpTrainClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._mode = None
self._training_op = None
class OpH5WriterBigDataset(Operator):
name = "H5 File Writer BigDataset"
category = "Output"
inputSlots = [
InputSlot(
"hdf5File"
), # Must be an already-open hdf5File (or group) for writing to
InputSlot("hdf5Path", stype="string"),
InputSlot("Image"),
InputSlot("CompressionEnabled", value=True),
InputSlot("BatchSize", optional=True)
]
outputSlots = [OutputSlot("WriteImage")]
loggingName = __name__ + ".OpH5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
def cleanUp(self):
super(OpH5WriterBigDataset, self).cleanUp()
# Discard the reference to the dataset, to ensure that hdf5 can close the file.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1, )
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["hdf5File"].value
hdf5Path = self.inputs["hdf5Path"].value
# On windows, there may be backslashes.
hdf5Path = hdf5Path.replace('\\', '/')
hdf5GroupName, datasetName = os.path.split(hdf5Path)
if hdf5GroupName == "":
g = self.f
else:
if hdf5GroupName in self.f:
g = self.f[hdf5GroupName]
else:
g = self.f.create_group(hdf5GroupName)
dataShape = self.Image.meta.shape
self.logger.info("Data shape: {}".format(dataShape))
dtype = self.Image.meta.dtype
if type(dtype) is numpy.dtype:
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
# Set up our chunk shape: Aim for a cube that's roughly 512k in size
dtypeBytes = dtype().nbytes
tagged_maxshape = self.Image.meta.getTaggedShape()
if 't' in tagged_maxshape:
# Assume that chunks should not span multiple t-slices,
# and channels are often handled separately, too.
tagged_maxshape['t'] = 1
if 'c' in tagged_maxshape:
tagged_maxshape['c'] = 1
self.chunkShape = determineBlockShape(tagged_maxshape.values(),
512000.0 / dtypeBytes)
if datasetName in g.keys():
del g[datasetName]
kwargs = {
'shape': dataShape,
'dtype': dtype,
'chunks': self.chunkShape
}
if self.CompressionEnabled.value:
kwargs[
'compression'] = 'gzip' # <-- Would be nice to use lzf compression here, but that is h5py-specific.
kwargs[
'compression_opts'] = 1 # <-- Optimize for speed, not disk space.
self.d = g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs['drange'] = self.Image.meta.drange
if self.Image.meta.display_mode is not None:
self.d.attrs['display_mode'] = self.Image.meta.display_mode
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
# Save the axistags as a dataset attribute
self.d.attrs['axistags'] = self.Image.meta.axistags.toJSON()
def handle_block_result(roi, data):
slicing = roiToSlice(*roi)
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=slicing)
else:
self.d[slicing] = data
batch_size = None
if self.BatchSize.ready():
batch_size = self.BatchSize.value
requester = BigRequestStreamer(self.Image,
roiFromShape(self.Image.meta.shape),
batchSize=batch_size)
requester.resultSignal.subscribe(handle_block_result)
requester.progressSignal.subscribe(self.progressSignal)
requester.execute()
# Be paranoid: Flush right now.
self.f.file.flush()
# We're finished.
result[0] = True
self.progressSignal(100)
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
class OpTikTorchTrainClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
nonzeroLabelBlocks = InputSlot(level=1) # Used only in the pixelwise case.
MaxLabel = InputSlot()
ModelSession = InputSlot()
BlockShape = InputSlot(level=1)
UpdatedModelSession = OutputSlot()
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def setupOutputs(self):
for slot in [self.Images, self.Labels]:
assert all(
[s.meta.getAxisKeys()[-1] == "c" for s in slot]
), f"This opearator assumes channel is the last axis. problem: {slot}"
self.ModelSession.meta.dtype = object
self.ModelSession.meta.shape = (1, )
def cleanUp(self):
self.progressSignal.clean()
super().cleanUp()
def _collect_blocks(self, image_slot, label_slot, block_slicings):
model = self.ModelSession.value
image_data_blocks = []
label_data_blocks = []
block_ids = []
for block_slicing in block_slicings:
# Get labels
block_label_roi = sliceToRoi(block_slicing, label_slot.meta.shape)
block_label_data = label_slot(*block_label_roi).wait()
bb_roi_within_block = numpy.array([[0] *
len(block_label_data.shape),
list(block_label_data.shape)])
block_label_bb_roi = bb_roi_within_block + block_label_roi[0]
# Ask for the halo needed by the classifier
axiskeys = image_slot.meta.getAxisKeys()
halo_shape = model.get_halo(axiskeys)
assert len(halo_shape) == len(block_label_roi[0])
assert halo_shape[
-1] == 0, "Didn't expect a non-zero halo for channel dimension."
# Expand block by halo, but keep clipped to image bounds
padded_label_roi, bb_roi_within_padded = enlargeRoiForHalo(
*block_label_bb_roi,
shape=label_slot.meta.shape,
sigma=halo_shape,
window=1,
return_result_roi=True)
# Copy labels to new array, which has size == bounding-box + halo
padded_label_data = numpy.zeros(
padded_label_roi[1] - padded_label_roi[0],
label_slot.meta.dtype)
padded_label_data[roiToSlice(
*bb_roi_within_padded)] = block_label_data[roiToSlice(
*bb_roi_within_block)]
padded_image_roi = numpy.array(padded_label_roi)
assert (padded_image_roi[:, -1] == [0, 1]).all()
num_channels = image_slot.meta.shape[-1]
padded_image_roi[:, -1] = [0, num_channels]
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
padded_image_data = numpy.asarray(
image_slot(*padded_image_roi).wait())
image_data_blocks.append(padded_image_data)
label_data_blocks.append(padded_label_data)
block_ids.append(
tuple(
int(block_label_bb_roi[0][i])
for i, key in enumerate(axiskeys) if key != "c"))
return image_data_blocks, label_data_blocks, block_ids
def execute(self, slot, subindex, roi, result):
model_session = self.ModelSession.value
result[0] = model_session
return result
def propagateDirty(self, slot, subindex, roi):
if slot == self.Labels:
if not self.ModelSession.ready():
return
try:
model_session = self.ModelSession.value
image_slot = self.Images[subindex]
label_slot = self.Labels[subindex]
# todo: get block shape in a less hacky way
block_shape = self.BlockShape[subindex].value
block_starts = getIntersectingBlocks(block_shape,
(roi.start, roi.stop))
label_shape = label_slot.meta.shape
axis_keys = label_slot.meta.getAxisKeys()
block_slicings = [[
slice(None) if axis == "c" else slice(dmax -
dblock, dmax) if
dstart + dblock > dmax else slice(dstart, dstart + dblock)
for dstart, dblock, dmax, axis in zip(
block_start, block_shape, label_shape, axis_keys)
] for block_start in block_starts]
image_blocks, label_blocks, block_ids = self._collect_blocks(
image_slot, label_slot, block_slicings)
axistags = self.Images[0].meta.axistags
model_session.update(image_blocks, label_blocks, axistags,
block_ids)
except Exception as e:
logger.error(e, exc_info=True)
else:
self.UpdatedModelSession.setDirty()
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def setupOutputs(self):
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1,)
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super( OpTrainPixelwiseClassifierBlocked, self ).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert isinstance(classifier_factory, LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
block_label_roi = sliceToRoi( block_slicing, image_slot.meta.shape )
block_image_roi = numpy.array( block_label_roi )
assert (block_image_roi[:, -1] == [0,1]).all()
num_channels = image_slot.meta.shape[-1]
block_image_roi[:, -1] = [0, num_channels]
# TODO: Compensate for the halo as specified by the classifier...
#axiskeys = image_slot.meta.getAxisKeys()
#halo_shape = classifier_factory.get_halo_shape(axiskeys)
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
block_label_data = numpy.asarray( label_slot(*block_label_roi).wait() )
block_image_data = numpy.asarray( image_slot(*block_image_roi).wait() )
label_data_blocks.append( block_label_data )
image_data_blocks.append( block_image_data )
classifier = classifier_factory.create_and_train_pixelwise( image_data_blocks, label_data_blocks )
assert isinstance(classifier, LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
result[0] = classifier
return result
def propagateDirty(self, slot, subindex, roi):
print 'classifier is dirty...'
self.Classifier.setDirty()
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked,
self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
# Normally, lane removal does not trigger a dirty notification.
# But in this case, if the lane contained any label data whatsoever,
# the classifier needs to be marked dirty.
# We know which slots contain (or contained) label data because they have
# been 'touched' at some point (they became dirty at some point).
self._touched_slots = set()
def handle_new_lane(multislot, index, newlength):
def handle_dirty_lane(slot, roi):
self._touched_slots.add(slot)
multislot[index].notifyDirty(handle_dirty_lane)
self.Labels.notifyInserted(handle_new_lane)
def handle_remove_lane(multislot, index, newlength):
# If the lane we're removing contained
# label data, then mark the downstream dirty
if multislot[index] in self._touched_slots:
self.Classifier.setDirty()
self._touched_slots.remove(multislot[index])
self.Labels.notifyRemove(handle_remove_lane)
def setupOutputs(self):
for slot in list(self.Images) + list(self.Labels):
assert slot.meta.getAxisKeys()[-1] == 'c', \
"This opearator assumes channel is the last axis."
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1, )
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super(OpTrainPixelwiseClassifierBlocked, self).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert issubclass(type(classifier_factory), LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(
self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
# Get labels
block_label_roi = sliceToRoi(block_slicing,
label_slot.meta.shape)
block_label_data = label_slot(*block_label_roi).wait()
# Shrink roi to bounding box of actual label pixels
bb_roi_within_block = nonzero_bounding_box(block_label_data)
block_label_bb_roi = bb_roi_within_block + block_label_roi[0]
# Double-check that there is at least 1 non-zero label in the block.
if (block_label_bb_roi[1] > block_label_bb_roi[0]).all():
# Ask for the halo needed by the classifier
axiskeys = image_slot.meta.getAxisKeys()
halo_shape = classifier_factory.get_halo_shape(axiskeys)
assert len(halo_shape) == len(block_label_roi[0])
assert halo_shape[
-1] == 0, "Didn't expect a non-zero halo for channel dimension."
# Expand block by halo, but keep clipped to image bounds
padded_label_roi, bb_roi_within_padded = enlargeRoiForHalo(
*block_label_bb_roi,
shape=label_slot.meta.shape,
sigma=halo_shape,
window=1,
return_result_roi=True)
# Copy labels to new array, which has size == bounding-box + halo
padded_label_data = numpy.zeros(
padded_label_roi[1] - padded_label_roi[0],
label_slot.meta.dtype)
padded_label_data[roiToSlice(
*bb_roi_within_padded)] = block_label_data[roiToSlice(
*bb_roi_within_block)]
padded_image_roi = numpy.array(padded_label_roi)
assert (padded_image_roi[:, -1] == [0, 1]).all()
num_channels = image_slot.meta.shape[-1]
padded_image_roi[:, -1] = [0, num_channels]
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
padded_image_data = numpy.asarray(
image_slot(*padded_image_roi).wait())
label_data_blocks.append(padded_label_data)
image_data_blocks.append(padded_image_data)
if len(image_data_blocks) == 0:
result[0] = None
else:
channel_names = self.Images[0].meta.channel_names
axistags = self.Images[0].meta.axistags
logger.debug("Training new pixelwise classifier: {}".format(
classifier_factory.description))
classifier = classifier_factory.create_and_train_pixelwise(
image_data_blocks, label_data_blocks, axistags, channel_names)
result[0] = classifier
if classifier is not None:
assert issubclass(type(classifier), LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
def propagateDirty(self, slot, subindex, roi):
self.Classifier.setDirty()
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked,
self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def setupOutputs(self):
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1, )
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super(OpTrainPixelwiseClassifierBlocked, self).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert isinstance(classifier_factory, LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(
self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
block_label_roi = sliceToRoi(block_slicing,
image_slot.meta.shape)
block_image_roi = numpy.array(block_label_roi)
assert (block_image_roi[:, -1] == [0, 1]).all()
num_channels = image_slot.meta.shape[-1]
block_image_roi[:, -1] = [0, num_channels]
# TODO: Compensate for the halo as specified by the classifier...
#axiskeys = image_slot.meta.getAxisKeys()
#halo_shape = classifier_factory.get_halo_shape(axiskeys)
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
block_label_data = numpy.asarray(
label_slot(*block_label_roi).wait())
block_image_data = numpy.asarray(
image_slot(*block_image_roi).wait())
label_data_blocks.append(block_label_data)
image_data_blocks.append(block_image_data)
classifier = classifier_factory.create_and_train_pixelwise(
image_data_blocks, label_data_blocks)
assert isinstance(classifier, LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
result[0] = classifier
return result
def propagateDirty(self, slot, subindex, roi):
print 'classifier is dirty...'
self.Classifier.setDirty()
class OpH5N5WriterBigDataset(Operator):
name = "H5 and N5 File Writer BigDataset"
category = "Output"
h5N5File = InputSlot() # Must be an already-open hdf5File/n5File (or group) for writing to
h5N5Path = InputSlot()
Image = InputSlot()
# h5py uses single-threaded gzip comression, which really slows down export.
CompressionEnabled = InputSlot(value=False)
BatchSize = InputSlot(optional=True)
WriteImage = OutputSlot()
loggingName = __name__ + ".OpH5N5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
def cleanUp(self):
super().cleanUp()
# Discard the reference to the dataset, to ensure that the file can be closed.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1,)
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["h5N5File"].value
h5N5Path = self.inputs["h5N5Path"].value
# On windows, there may be backslashes.
h5N5Path = h5N5Path.replace("\\", "/")
h5N5GroupName, datasetName = os.path.split(h5N5Path)
if h5N5GroupName == "":
g = self.f
else:
if h5N5GroupName in self.f:
g = self.f[h5N5GroupName]
else:
g = self.f.create_group(h5N5GroupName)
dataShape = self.Image.meta.shape
self.logger.info(f"Data shape: {dataShape}")
dtype = self.Image.meta.dtype
if isinstance(dtype, numpy.dtype):
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
# Set up our chunk shape: Aim for a cube that's roughly 512k in size
dtypeBytes = dtype().nbytes
tagged_maxshape = self.Image.meta.getTaggedShape()
if "t" in tagged_maxshape:
# Assume that chunks should not span multiple t-slices,
# and channels are often handled separately, too.
tagged_maxshape["t"] = 1
if "c" in tagged_maxshape:
tagged_maxshape["c"] = 1
self.chunkShape = determineBlockShape(list(tagged_maxshape.values()), 512_000.0 / dtypeBytes)
if datasetName in list(g.keys()):
del g[datasetName]
kwargs = {"shape": dataShape, "dtype": dtype, "chunks": self.chunkShape}
if self.CompressionEnabled.value:
kwargs["compression"] = "gzip" # <-- Would be nice to use lzf compression here, but that is h5py-specific.
if isinstance(self.f, h5py.File):
kwargs["compression_opts"] = 1 # <-- Optimize for speed, not disk space.
else: # z5py has uses different names here
kwargs["level"] = 1 # <-- Optimize for speed, not disk space.
else:
if isinstance(self.f, z5py.N5File): # n5 uses gzip level 5 as default compression.
kwargs["compression"] = "raw"
self.d = g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs["drange"] = self.Image.meta.drange
if self.Image.meta.display_mode is not None:
self.d.attrs["display_mode"] = self.Image.meta.display_mode
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
# Save the axistags as a dataset attribute
self.d.attrs["axistags"] = self.Image.meta.axistags.toJSON()
def handle_block_result(roi, data):
slicing = roiToSlice(*roi)
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=slicing)
else:
self.d[slicing] = data
batch_size = None
if self.BatchSize.ready():
batch_size = self.BatchSize.value
requester = BigRequestStreamer(self.Image, roiFromShape(self.Image.meta.shape), batchSize=batch_size)
requester.resultSignal.subscribe(handle_block_result)
requester.progressSignal.subscribe(self.progressSignal)
requester.execute()
# Be paranoid: Flush right now.
if isinstance(self.f, h5py.File):
self.f.file.flush() # not available in z5py
# We're finished.
result[0] = True
self.progressSignal(100)
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
def __init__(self, *args, **kwargs):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
class OpTrainPixelwiseClassifierBlocked(Operator):
Images = InputSlot(level=1)
Labels = InputSlot(level=1)
ClassifierFactory = InputSlot()
nonzeroLabelBlocks = InputSlot(level=1)
MaxLabel = InputSlot()
Classifier = OutputSlot()
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def setupOutputs(self):
for slot in list(self.Images) + list(self.Labels):
assert slot.meta.getAxisKeys()[-1] == 'c', \
"This opearator assumes channel is the last axis."
self.Classifier.meta.dtype = object
self.Classifier.meta.shape = (1,)
# Special metadata for downstream operators using the classifier
self.Classifier.meta.classifier_factory = self.ClassifierFactory.value
def cleanUp(self):
self.progressSignal.clean()
super( OpTrainPixelwiseClassifierBlocked, self ).cleanUp()
def execute(self, slot, subindex, roi, result):
classifier_factory = self.ClassifierFactory.value
assert issubclass(type(classifier_factory), LazyflowPixelwiseClassifierFactoryABC), \
"Factory is of type {}, which does not satisfy the LazyflowPixelwiseClassifierFactoryABC interface."\
"".format( type(classifier_factory) )
# Accumulate all non-zero blocks of each image into lists
label_data_blocks = []
image_data_blocks = []
for image_slot, label_slot, nonzero_block_slot in zip(self.Images, self.Labels, self.nonzeroLabelBlocks):
block_slicings = nonzero_block_slot.value
for block_slicing in block_slicings:
block_label_roi = sliceToRoi( block_slicing, image_slot.meta.shape )
# Ask for the halo needed by the classifier
axiskeys = image_slot.meta.getAxisKeys()
halo_shape = classifier_factory.get_halo_shape(axiskeys)
assert len(halo_shape) == len( block_label_roi[0] )
assert halo_shape[-1] == 0, "Didn't expect a non-zero halo for channel dimension."
# Expand block by halo, then clip to image bounds
block_label_roi = numpy.array( block_label_roi )
block_label_roi[0] -= halo_shape
block_label_roi[1] += halo_shape
block_label_roi = getIntersection( block_label_roi, roiFromShape(image_slot.meta.shape) )
block_image_roi = numpy.array( block_label_roi )
assert (block_image_roi[:, -1] == [0,1]).all()
num_channels = image_slot.meta.shape[-1]
block_image_roi[:, -1] = [0, num_channels]
# Ensure the results are plain ndarray, not VigraArray,
# which some classifiers might have trouble with.
block_label_data = numpy.asarray( label_slot(*block_label_roi).wait() )
block_image_data = numpy.asarray( image_slot(*block_image_roi).wait() )
label_data_blocks.append( block_label_data )
image_data_blocks.append( block_image_data )
classifier = classifier_factory.create_and_train_pixelwise( image_data_blocks, label_data_blocks )
assert issubclass(type(classifier), LazyflowPixelwiseClassifierABC), \
"Classifier is of type {}, which does not satisfy the LazyflowPixelwiseClassifierABC interface."\
"".format( type(classifier) )
result[0] = classifier
return result
def propagateDirty(self, slot, subindex, roi):
self.Classifier.setDirty()
class OpH5N5WriterBigDataset(Operator):
name = "H5 and N5 File Writer BigDataset"
category = "Output"
h5N5File = InputSlot() # Must be an already-open hdf5File/n5File (or group) for writing to
h5N5Path = InputSlot()
Image = InputSlot()
# h5py uses single-threaded gzip comression, which really slows down export.
CompressionEnabled = InputSlot(value=False)
BatchSize = InputSlot(optional=True)
WriteImage = OutputSlot()
loggingName = __name__ + ".OpH5N5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(
self,
h5N5File=None,
h5N5Path=None,
Image=None,
BatchSize: int = None,
CompressionEnabled: bool = None,
*args,
**kwargs,
):
super().__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
self.h5N5File.setOrConnectIfAvailable(h5N5File)
self.h5N5Path.setOrConnectIfAvailable(h5N5Path)
self.Image.setOrConnectIfAvailable(Image)
self.BatchSize.setOrConnectIfAvailable(BatchSize)
self.CompressionEnabled.setOrConnectIfAvailable(CompressionEnabled)
def cleanUp(self):
super().cleanUp()
# Discard the reference to the dataset, to ensure that the file can be closed.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1,)
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["h5N5File"].value
h5N5Path = self.inputs["h5N5Path"].value
# On windows, there may be backslashes.
h5N5Path = h5N5Path.replace("\\", "/")
h5N5GroupName, datasetName = os.path.split(h5N5Path)
if h5N5GroupName == "":
g = self.f
else:
if h5N5GroupName in self.f:
g = self.f[h5N5GroupName]
else:
g = self.f.create_group(h5N5GroupName)
dataShape = self.Image.meta.shape
self.logger.info(f"Data shape: {dataShape}")
dtype = self.Image.meta.dtype
if isinstance(dtype, numpy.dtype):
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
# Set up our chunk shape: Aim for a cube that's roughly 512k in size
dtypeBytes = dtype().nbytes
tagged_maxshape = self.Image.meta.getTaggedShape()
if "t" in tagged_maxshape:
# Assume that chunks should not span multiple t-slices,
# and channels are often handled separately, too.
tagged_maxshape["t"] = 1
if "c" in tagged_maxshape:
tagged_maxshape["c"] = 1
self.chunkShape = determineBlockShape(list(tagged_maxshape.values()), 512_000.0 / dtypeBytes)
if datasetName in list(g.keys()):
del g[datasetName]
kwargs = {"shape": dataShape, "dtype": dtype, "chunks": self.chunkShape}
if self.CompressionEnabled.value:
kwargs["compression"] = "gzip" # <-- Would be nice to use lzf compression here, but that is h5py-specific.
if isinstance(self.f, h5py.File):
kwargs["compression_opts"] = 1 # <-- Optimize for speed, not disk space.
else: # z5py has uses different names here
kwargs["level"] = 1 # <-- Optimize for speed, not disk space.
else:
if isinstance(self.f, z5py.N5File): # n5 uses gzip level 5 as default compression.
kwargs["compression"] = "raw"
self.d = g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs["drange"] = self.Image.meta.drange
if self.Image.meta.display_mode is not None:
self.d.attrs["display_mode"] = self.Image.meta.display_mode
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
# Save the axistags as a dataset attribute
self.d.attrs["axistags"] = self.Image.meta.axistags.toJSON()
if isinstance(self.d, h5py.Dataset):
for index, tag in enumerate(self.Image.meta.axistags):
self.d.dims[index].label = tag.key
else: # if n5 dataset, apply neuroglancer's axes tags convention
self.d.attrs["axes"] = "".join(tag.key for tag in self.Image.meta.axistags)[::-1]
drange = self.Image.meta.get("drange")
if drange:
self.d.attrs["drange"] = drange
def handle_block_result(roi, data):
slicing = roiToSlice(*roi)
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=slicing)
else:
self.d[slicing] = data
batch_size = None
if self.BatchSize.ready():
batch_size = self.BatchSize.value
requester = BigRequestStreamer(self.Image, roiFromShape(self.Image.meta.shape), batchSize=batch_size)
requester.resultSignal.subscribe(handle_block_result)
requester.progressSignal.subscribe(self.progressSignal)
requester.execute()
# Be paranoid: Flush right now.
if isinstance(self.f, h5py.File):
self.f.file.flush() # not available in z5py
# We're finished.
result[0] = True
self.progressSignal(100)
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
def __init__(self, *args, **kwargs):
super(OpTaskWorker, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self._primaryBlockwiseFileset = None
class OpH5WriterBigDataset(Operator):
name = "H5 File Writer BigDataset"
category = "Output"
inputSlots = [InputSlot("hdf5File"), # Must be an already-open hdf5File (or group) for writing to
InputSlot("hdf5Path", stype = "string"),
InputSlot("Image"),
InputSlot("CompressionEnabled", value=True)]
outputSlots = [OutputSlot("WriteImage")]
loggingName = __name__ + ".OpH5WriterBigDataset"
logger = logging.getLogger(loggingName)
traceLogger = logging.getLogger("TRACE." + loggingName)
def __init__(self, *args, **kwargs):
super(OpH5WriterBigDataset, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
self.d = None
self.f = None
def cleanUp(self):
super( OpH5WriterBigDataset, self ).cleanUp()
# Discard the reference to the dataset, to ensure that hdf5 can close the file.
self.d = None
self.f = None
self.progressSignal.clean()
def setupOutputs(self):
self.outputs["WriteImage"].meta.shape = (1,)
self.outputs["WriteImage"].meta.dtype = object
self.f = self.inputs["hdf5File"].value
hdf5Path = self.inputs["hdf5Path"].value
# On windows, there may be backslashes.
hdf5Path = hdf5Path.replace('\\', '/')
hdf5GroupName, datasetName = os.path.split(hdf5Path)
if hdf5GroupName == "":
g = self.f
else:
if hdf5GroupName in self.f:
g = self.f[hdf5GroupName]
else:
g = self.f.create_group(hdf5GroupName)
dataShape=self.Image.meta.shape
self.logger.info( "Data shape: {}".format(dataShape))
dtype = self.Image.meta.dtype
if type(dtype) is numpy.dtype:
# Make sure we're dealing with a type (e.g. numpy.float64),
# not a numpy.dtype
dtype = dtype.type
# Set up our chunk shape: Aim for a cube that's roughly 512k in size
dtypeBytes = dtype().nbytes
tagged_maxshape = self.Image.meta.getTaggedShape()
if 't' in tagged_maxshape:
# Assume that chunks should not span multiple t-slices
tagged_maxshape['t'] = 1
self.chunkShape = determineBlockShape( tagged_maxshape.values(), 512000.0 / dtypeBytes )
if datasetName in g.keys():
del g[datasetName]
kwargs = { 'shape' : dataShape, 'dtype' : dtype,
'chunks' : self.chunkShape }
if self.CompressionEnabled.value:
kwargs['compression'] = 'gzip' # <-- Would be nice to use lzf compression here, but that is h5py-specific.
kwargs['compression_opts'] = 1 # <-- Optimize for speed, not disk space.
self.d=g.create_dataset(datasetName, **kwargs)
if self.Image.meta.drange is not None:
self.d.attrs['drange'] = self.Image.meta.drange
def execute(self, slot, subindex, rroi, result):
self.progressSignal(0)
# Save the axistags as a dataset attribute
self.d.attrs['axistags'] = self.Image.meta.axistags.toJSON()
def handle_block_result(roi, data):
slicing = roiToSlice(*roi)
if data.flags.c_contiguous:
self.d.write_direct(data.view(numpy.ndarray), dest_sel=slicing)
else:
self.d[slicing] = data
requester = BigRequestStreamer( self.Image, roiFromShape( self.Image.meta.shape ) )
requester.resultSignal.subscribe( handle_block_result )
requester.progressSignal.subscribe( self.progressSignal )
requester.execute()
# Be paranoid: Flush right now.
self.f.file.flush()
# We're finished.
result[0] = True
self.progressSignal(100)
def propagateDirty(self, slot, subindex, roi):
# The output from this operator isn't generally connected to other operators.
# If someone is using it that way, we'll assume that the user wants to know that
# the input image has become dirty and may need to be written to disk again.
self.WriteImage.setDirty(slice(None))
def __init__(self, *args, **kwargs):
super(OpTrainPixelwiseClassifierBlocked, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
def __init__(self, *args, **kwargs):
super(OpTrainClassifierFromFeatureVectorsAndSupervoxelMask,
self).__init__(*args, **kwargs)
self.trainingCompleteSignal = OrderedSignal()
def __init__(self, *args, **kwargs):
super(OpStackWriter, self).__init__(*args, **kwargs)
self.progressSignal = OrderedSignal()
