def run_export(self):
role_names = self.parentApplet.dataSelectionApplet.topLevelOperator.DatasetRoles.value
# Prepare file lists in an OrderedDict
role_path_dict = OrderedDict()
role_path_dict[0] = BatchProcessingGui.get_all_item_strings(
self.list_widgets[0])
num_datasets = len(role_path_dict[0])
for role_index, list_widget in enumerate(self.list_widgets[1:],
start=1):
role_path_dict[
role_index] = BatchProcessingGui.get_all_item_strings(
self.list_widgets[role_index])
assert len(role_path_dict[role_index]) <= num_datasets, \
"Too many files given for role: '{}'".format( role_names[role_index] )
if len(role_path_dict[role_index]) < num_datasets:
role_path_dict[role_index] += [None] * (
num_datasets - len(role_path_dict[role_index]))
# Run the export in a separate thread
export_req = Request(
partial(self.parentApplet.run_export, role_path_dict))
export_req.notify_failed(self.handle_batch_processing_failure)
export_req.notify_finished(self.handle_batch_processing_finished)
export_req.notify_cancelled(self.handle_batch_processing_cancelled)
self.export_req = export_req
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested.emit()
self.cancel_button.setVisible(True)
self.run_button.setEnabled(False)
# Start the export
export_req.submit()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('grayscale', help='example: my-grayscale.h5/volume')
parser.add_argument('classifier', help='example: my-file.h5/forest')
parser.add_argument('filter_specs',
help='json file containing filter list')
parser.add_argument('output_path',
help='example: my-predictions.h5/volume')
parser.add_argument('--compute-blockwise',
help='Compute blockwise instead of as a whole',
action='store_true')
parser.add_argument('--thread-count',
help='The threadpool size',
default=0,
type=int)
args = parser.parse_args()
# Show log messages on the console.
logger.setLevel(logging.INFO)
logger.addHandler(logging.StreamHandler(sys.stdout))
Request.reset_thread_pool(args.thread_count)
load_and_predict(args.grayscale, args.classifier, args.filter_specs,
args.output_path, args.compute_blockwise)
logger.info("DONE.")
def predict_probabilities(self, X):
logger.debug( "Predicting with parallel vigra RF" )
X = numpy.asarray(X, dtype=numpy.float32)
# As each forest completes, aggregate results in a shared array.
# (Must put in a list so we can update it in this closure.)
total_predictions = [None]
prediction_lock = RequestLock()
def update_predictions(forest, forest_predictions):
forest_predictions *= forest.treeCount()
with prediction_lock:
if total_predictions[0] is None:
total_predictions[0] = forest_predictions
else:
total_predictions[0] += forest_predictions
# Create a request for each forest
pool = RequestPool()
for forest in self._forests:
req = Request( partial( forest.predictProbabilities, X ) )
req.notify_finished( partial(update_predictions, forest) )
pool.add( req )
del req
pool.wait()
total_predictions[0] /= self._num_trees
return total_predictions[0]
def test(s):
req = Request(someWork)
req.notify(callback)
req.wait()
time.sleep(0.001)
print s
return s
def predict_probabilities(self, X):
logger.debug( "Predicting with parallel vigra RF" )
X = numpy.asarray(X, dtype=numpy.float32)
# As each forest completes, aggregate results in a shared array.
# (Must put in a list so we can update it in this closure.)
total_predictions = [None]
prediction_lock = RequestLock()
def update_predictions(forest, forest_predictions):
forest_predictions *= forest.treeCount()
with prediction_lock:
if total_predictions[0] is None:
total_predictions[0] = forest_predictions
else:
total_predictions[0] += forest_predictions
# Create a request for each forest
pool = RequestPool()
for forest in self._forests:
req = Request( partial( forest.predictProbabilities, X ) )
req.notify_finished( partial(update_predictions, forest) )
pool.add( req )
del req
pool.wait()
total_predictions[0] /= self._num_trees
return total_predictions[0]
def test_basic(self):
def someWork():
time.sleep(0.001)
#print "producer finished"
def callback(s):
pass
def test(s):
req = Request(someWork)
req.notify(callback)
req.wait()
time.sleep(0.001)
print s
return s
req = Request( test, s = "hallo !")
req.notify(callback)
assert req.wait() == "hallo !"
requests = []
for i in range(10):
req = Request( test, s = "hallo %d" %i)
requests.append(req)
for r in requests:
r.wait()
def test_withH5Py(self):
"""
We have suspicions that greenlet and h5py don't interact well with eachother.
This tests basic functionality.
TODO: Expand it for better coverage.
"""
maxDepth = 5
maxBreadth = 10
filename = 'requestTest.h5'
h5File = h5py.File( filename, 'w' )
dataset = h5File.create_dataset( 'test/data', data=numpy.zeros( (maxDepth, maxBreadth), dtype=int ))
def writeToH5Py(result, index, req):
dataset[index] += 1
# This closure randomly chooses to either (a) return immediately or (b) fire off more work
def someWork(depth, force=False, i=0):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.5):
requests = []
for i in range(maxBreadth):
req = Request(someWork, depth=depth-1, i=i)
req.notify(writeToH5Py, index=(depth-1, i), req=req)
requests.append(req)
for r in requests:
r.wait()
req = Request(someWork, depth=maxDepth, force=True)
req.wait()
h5File.close()
print "finished testWithH5Py"
os.remove(filename)
def test_withH5Py(self):
"""
We have suspicions that greenlet and h5py don't interact well with eachother.
This tests basic functionality.
TODO: Expand it for better coverage.
"""
maxDepth = 5
maxBreadth = 10
filename = 'requestTest.h5'
h5File = h5py.File( filename, 'w' )
dataset = h5File.create_dataset( 'test/data', data=numpy.zeros( (maxDepth, maxBreadth), dtype=int ))
def writeToH5Py(result, index, req):
dataset[index] += 1
# This closure randomly chooses to either (a) return immediately or (b) fire off more work
def someWork(depth, force=False, i=0):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.5):
requests = []
for i in range(maxBreadth):
req = Request(someWork, depth=depth-1, i=i)
req.notify(writeToH5Py, index=(depth-1, i), req=req)
requests.append(req)
for r in requests:
r.wait()
req = Request(someWork, depth=maxDepth, force=True)
req.wait()
h5File.close()
print "finished testWithH5Py"
os.remove(filename)
def predict_probabilities(self, X):
logger.debug( "Predicting with parallel vigra RF" )
X = numpy.asarray(X, dtype=numpy.float32)
assert X.ndim == 2
if self._feature_names is not None:
# For some reason, vigra doesn't seem to check this for us...
assert X.shape[1] == len(self._feature_names), \
"Feature count ({}) doesn't match the training feature count ({}).\n"\
"Expected features: {}".format( X.shape[1], len(self._feature_names), self._feature_names )
# As each forest completes, aggregate results in a shared array.
# (Must put in a list so we can update it in this closure.)
total_predictions = [None]
prediction_lock = RequestLock()
def update_predictions(forest, forest_predictions):
forest_predictions *= forest.treeCount()
with prediction_lock:
if total_predictions[0] is None:
total_predictions[0] = forest_predictions
else:
total_predictions[0] += forest_predictions
# Create a request for each forest
pool = RequestPool()
for forest in self._forests:
req = Request( partial( forest.predictProbabilities, X ) )
req.notify_finished( partial(update_predictions, forest) )
pool.add( req )
del req
pool.wait()
total_predictions[0] /= self._num_trees
return total_predictions[0]
def exportObjectCounts(self):
opCounting = self.parentApplet.opCounting
export_filepath = QFileDialog.getSaveFileName(parent=self, caption="Exported Object Counts", filter="*.csv")
if not export_filepath:
return
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested.emit()
def _exportObjectCounts():
num_files = len(self.topLevelOperator.RawDatasetInfo)
with open(export_filepath, 'w') as export_file:
for lane_index, (info_slot, sum_slot) in enumerate(zip(self.topLevelOperator.RawDatasetInfo, opCounting.OutputSum)):
self.parentApplet.progressSignal.emit(100.0*lane_index/num_files)
nickname = info_slot.value.nickname
object_count = sum_slot[:].wait()[0]
export_file.write(nickname + "," + str(object_count) + "\n")
self.parentApplet.busy = False
self.parentApplet.progressSignal.emit(100)
self.parentApplet.appletStateUpdateRequested.emit()
req = Request(_exportObjectCounts)
req.notify_failed( self.handleFailedObjectCountExport )
req.submit()
def predict_probabilities(self, X):
logger.debug("Predicting with parallel vigra RF")
X = numpy.asarray(X, dtype=numpy.float32)
assert X.ndim == 2
if self._feature_names is not None:
# For some reason, vigra doesn't seem to check this for us...
assert X.shape[1] == len(self._feature_names), \
"Feature count doesn't match the training data."
# As each forest completes, aggregate results in a shared array.
# (Must put in a list so we can update it in this closure.)
total_predictions = [None]
prediction_lock = RequestLock()
def update_predictions(forest, forest_predictions):
forest_predictions *= forest.treeCount()
with prediction_lock:
if total_predictions[0] is None:
total_predictions[0] = forest_predictions
else:
total_predictions[0] += forest_predictions
# Create a request for each forest
pool = RequestPool()
for forest in self._forests:
req = Request(partial(forest.predictProbabilities, X))
req.notify_finished(partial(update_predictions, forest))
pool.add(req)
del req
pool.wait()
total_predictions[0] /= self._num_trees
return total_predictions[0]
def run_export(self):
role_names = self.parentApplet.dataSelectionApplet.role_names
# Prepare file lists in an OrderedDict
role_inputs = {
role_name: self._data_role_widgets[role_name].filepaths
for role_name in role_names
}
if all(len(role_inp) == 0 for role_inp in role_inputs.values()):
return
# Run the export in a separate thread
lane_configs = self.parentApplet.dataSelectionApplet.create_lane_configs(
role_inputs=role_inputs)
export_req = Request(
partial(self.parentApplet.run_export, lane_configs=lane_configs))
export_req.notify_failed(self.handle_batch_processing_failure)
export_req.notify_finished(self.handle_batch_processing_finished)
export_req.notify_cancelled(self.handle_batch_processing_cancelled)
self.export_req = export_req
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested()
self.cancel_button.setVisible(True)
self.run_button.setEnabled(False)
# Start the export
export_req.submit()
def create_and_train(self, X, y):
logger.debug( "Training parallel vigra RF" )
# Save for future reference
known_labels = numpy.unique(y)
X = numpy.asarray(X, numpy.float32)
y = numpy.asarray(y, numpy.uint32)
if y.ndim == 1:
y = y[:, numpy.newaxis]
assert X.ndim == 2
assert len(X) == len(y)
# Create N forests
forests = []
for _ in range(self._num_forests):
forest = vigra.learning.RandomForest(self._trees_per_forest, **self._kwargs)
forests.append( forest )
# Train them all in parallel
oobs = [None] * len(forests)
pool = RequestPool()
for i, forest in enumerate(forests):
req = Request( partial(forest.learnRF, X, y) )
# save the oobs
req.notify_finished( partial( oobs.__setitem__, i ) )
pool.add( req )
pool.wait()
return ParallelVigraRfLazyflowClassifier( forests, oobs, known_labels )
def submit_to_threadpool(fn, priority):
if USE_LAZYFLOW_THREADPOOL:
# Tiling requests are less prioritized than most requests.
root_priority = [1] + list(priority)
req = Request(fn, root_priority)
req.submit()
else:
get_render_pool().submit(fn, priority)
def lots_of_work():
requests = []
for i in range(mcount):
req = Request(functools.partial(empty_func, b=11))
req.submit()
for r in requests:
r.wait()
def lots_of_work():
requests = []
for i in range(mcount):
req = Request(functools.partial(empty_func, b = 11))
req.submit()
for r in requests:
r.wait()
def teardown_method(self, method):
# reset cleanup frequency to sane value
# reset max memory
Memory.setAvailableRamCaches(-1)
mgr = CacheMemoryManager()
mgr.setRefreshInterval(default_refresh_interval)
mgr.enable()
Request.reset_thread_pool()
def submit_to_threadpool(fn, priority):
if USE_LAZYFLOW_THREADPOOL:
# Tiling requests are less prioritized than most requests.
root_priority = [1] + list(priority)
req = Request(fn, root_priority)
req.submit()
else:
get_render_pool().submit(fn, priority)
def teardown_method(self, method):
# reset cleanup frequency to sane value
# reset max memory
Memory.setAvailableRamCaches(-1)
mgr = _CacheMemoryManager()
mgr.setRefreshInterval(default_refresh_interval)
mgr.enable()
Request.reset_thread_pool()
def _exportMeshes(self, object_names: List[str], obj_filepaths: List[str]) -> Request:
"""Save objects in the mst to .obj files
Args:
object_names: Names of the objects in the mst
obj_filepaths: One path for each object in object_names
Returns:
Returns the request object, used in testing
"""
def get_label_volume_from_mst(mst, object_name):
object_supervoxels = mst.object_lut[object_name]
object_lut = numpy.zeros(mst.nodeNum+1, dtype=numpy.int32)
object_lut[object_supervoxels] = 1
supervoxel_volume = mst.supervoxelUint32
object_volume = object_lut[supervoxel_volume]
return object_volume
mst = self.topLevelOperatorView.MST.value
def exportMeshes(object_names, obj_filepaths):
n_objects = len(object_names)
progress_update = 100 / n_objects
try:
for obj, obj_path, obj_n in zip(object_names, obj_filepaths, range(n_objects)):
object_volume = get_label_volume_from_mst(mst, obj)
unique_ids = len(numpy.unique(object_volume))
if unique_ids <= 1:
logger.info(f"No voxels found for {obj}, skipping")
continue
elif unique_ids > 2:
logger.info(f"Supervoxel segmentation not unique for {obj}, skipping, got {unique_ids}")
continue
logger.info(f"Generating mesh for {obj}")
_, mesh_data = list(labeling_to_mesh(object_volume, [1]))[0]
self.parentApplet.progressSignal((obj_n + .5) * progress_update)
logger.info(f"Mesh generation for {obj} complete.")
logger.info(f"Saving mesh for {obj} to {obj_path}")
mesh_to_obj(mesh_data, obj_path, obj)
self.parentApplet.progressSignal((obj_n + 1) * progress_update)
finally:
self.parentApplet.busy = False
self.parentApplet.progressSignal(100)
self.parentApplet.appletStateUpdateRequested()
self.parentApplet.busy = True
self.parentApplet.progressSignal(-1)
self.parentApplet.appletStateUpdateRequested()
req = Request(partial(exportMeshes, object_names, obj_filepaths))
req.submit()
return req
def test_RequestLock():
assert Request.global_thread_pool.num_workers > 0, \
"This test must be used with the real threadpool."
lockA = RequestLock()
lockB = RequestLock()
def log_request_system_status():
status = (
"*************************\n" +
'lockA.pending: {}\n'.format(len(lockA._pendingRequests)) +
'lockB.pending: {}\n'.format(len(lockB._pendingRequests))
#+ "suspended Requests: {}\n".format( len(Request.global_suspend_set) )
+ "global job queue: {}\n".format(
len(Request.global_thread_pool.unassigned_tasks)))
for worker in Request.global_thread_pool.workers:
status += "{} queued tasks: {}\n".format(worker.name,
len(worker.job_queue))
status += "*****************************************************"
logger.debug(status)
running = [True]
def periodic_status():
while running[0]:
time.sleep(0.5)
log_request_system_status()
# Uncomment these lines to print periodic status while the test runs...
status_thread = threading.Thread(target=periodic_status)
status_thread.daemon = True
status_thread.start()
try:
_impl_test_lock(lockA, lockB, Request, 1000)
except:
log_request_system_status()
running[0] = False
status_thread.join()
global paused
paused = False
Request.reset_thread_pool(Request.global_thread_pool.num_workers)
if lockA.locked():
lockA.release()
if lockB.locked():
lockB.release()
raise
log_request_system_status()
running[0] = False
status_thread.join()
def test_pause_unpause(self):
handlerCounter = [0]
handlerLock = threading.Lock()
def completionHandler( result, req ):
handlerLock.acquire()
handlerCounter[0] += 1
handlerLock.release()
requestCounter = [0]
requestLock = threading.Lock()
allRequests = []
# This closure randomly chooses to either (a) return immediately or (b) fire off more work
def someWork(depth, force=False, i=-1):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.8):
requests = []
for i in range(10):
req = Request(someWork, depth=depth-1, i=i)
req.notify(completionHandler, req=req)
requests.append(req)
allRequests.append(req)
requestLock.acquire()
requestCounter[0] += 1
requestLock.release()
for r in requests:
r.wait()
req = Request(someWork, depth=6, force=True)
def blubb(req):
pass
req.notify(blubb)
global_thread_pool.pause()
req2 = Request(someWork, depth=6, force=True)
req2.notify(blubb)
global_thread_pool.unpause()
assert req2.finished == False
assert req.finished
req.wait()
# Handler should have been called once for each request we fired
assert handlerCounter[0] == requestCounter[0]
print "finished pause_unpause"
for r in allRequests:
assert r.finished
print "waited for all subrequests"
def run_export(self):
role_names = self.parentApplet.dataSelectionApplet.topLevelOperator.DatasetRoles.value
# Prepare file lists in an OrderedDict
role_path_dict = OrderedDict()
role_path_dict[0] = BatchProcessingGui.get_all_item_strings(self.list_widgets[0])
num_datasets = len(role_path_dict[0])
for role_index, list_widget in enumerate(self.list_widgets[1:], start=1):
role_path_dict[role_index] = BatchProcessingGui.get_all_item_strings(self.list_widgets[role_index])
assert len(role_path_dict[role_index]) <= num_datasets, \
"Too many files given for role: '{}'".format( role_names[role_index] )
if len(role_path_dict[role_index]) < num_datasets:
role_path_dict[role_index] += [None] * (num_datasets-len(role_path_dict[role_index]))
# Run the export in a separate thread
export_req = Request(partial(self.parentApplet.run_export, role_path_dict))
export_req.notify_failed(self.handle_batch_processing_failure)
export_req.notify_finished(self.handle_batch_processing_finished)
export_req.notify_cancelled(self.handle_batch_processing_cancelled)
self.export_req = export_req
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested()
self.cancel_button.setVisible(True)
self.run_button.setEnabled(False)
# Start the export
export_req.submit()
def _triggerTableUpdate(self):
# Check that object area is included in selected features
featureNames = self.topLevelOperatorView.SelectedFeatures.value
if 'Standard Object Features' not in featureNames or 'Count' not in featureNames['Standard Object Features']:
box = QMessageBox(QMessageBox.Warning,
'Warning',
'Object area is not a selected feature. Please select this feature on: \"Standard Object Features > Shape > Size in pixels\"',
QMessageBox.NoButton,
self)
box.show()
return
# Clear table
self.table.clearContents()
self.table.setRowCount(0)
self.table.setSortingEnabled(False)
self.progressBar.show()
self.computeButton.setEnabled(False)
def compute_features_for_frame(tIndex, t, features):
# Compute features and labels (called in parallel from request pool)
roi = [slice(None) for i in range(len(self.topLevelOperatorView.LabelImages.meta.shape))]
roi[tIndex] = slice(t, t+1)
roi = tuple(roi)
frame = self.topLevelOperatorView.SegmentationImages(roi).wait()
frame = frame.squeeze().astype(numpy.uint32, copy=False)
# Dirty trick: We don't care what we're passing here for the 'image' parameter,
# but vigra insists that we pass *something*, so we'll cast the label image as float32.
features[t] = vigra.analysis.extractRegionFeatures(frame.view(numpy.float32),
frame,
['Count'],
ignoreLabel=0)
tIndex = self.topLevelOperatorView.SegmentationImages.meta.axistags.index('t')
tMax = self.topLevelOperatorView.SegmentationImages.meta.shape[tIndex]
features = {}
labels = {}
def compute_all_features():
# Compute features in parallel
pool = RequestPool()
for t in range(tMax):
pool.add( Request( partial(compute_features_for_frame, tIndex, t, features) ) )
pool.wait()
# Compute labels
labels = self.topLevelOperatorView.LabelInputs([]).wait()
req = Request(compute_all_features)
req.notify_finished( partial(self._populateTable, features, labels) )
req.submit()
def addStack(self, roleIndex, laneIndex):
"""
The user clicked the "Import Stack Files" button.
"""
stackDlg = StackFileSelectionWidget(self)
stackDlg.exec_()
if stackDlg.result() != QDialog.Accepted :
return
files = stackDlg.selectedFiles
if len(files) == 0:
return
info = DatasetInfo()
info.filePath = "//".join( files )
prefix = os.path.commonprefix(files)
info.nickname = PathComponents(prefix).filenameBase
# Add an underscore for each wildcard digit
num_wildcards = len(files[-1]) - len(prefix) - len( os.path.splitext(files[-1])[1] )
info.nickname += "_"*num_wildcards
# Allow labels by default if this gui isn't being used for batch data.
info.allowLabels = ( self.guiMode == GuiMode.Normal )
info.fromstack = True
originalNumLanes = len(self.topLevelOperator.DatasetGroup)
if laneIndex is None:
laneIndex = len(self.topLevelOperator.DatasetGroup)
if len(self.topLevelOperator.DatasetGroup) < laneIndex+1:
self.topLevelOperator.DatasetGroup.resize(laneIndex+1)
def importStack():
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested.emit()
# Serializer will update the operator for us, which will propagate to the GUI.
try:
self.serializer.importStackAsLocalDataset( info )
try:
self.topLevelOperator.DatasetGroup[laneIndex][roleIndex].setValue(info)
except DatasetConstraintError as ex:
# Give the user a chance to repair the problem.
filename = files[0] + "\n...\n" + files[-1]
return_val = [False]
self.handleDatasetConstraintError( info, filename, ex, roleIndex, laneIndex, return_val )
if not return_val[0]:
# Not successfully repaired. Roll back the changes and give up.
self.topLevelOperator.DatasetGroup.resize(originalNumLanes)
finally:
self.parentApplet.busy = False
self.parentApplet.appletStateUpdateRequested.emit()
req = Request( importStack )
req.notify_failed( partial(self.handleFailedStackLoad, files, originalNumLanes ) )
req.submit()
def setup(self):
super(SkeletonAssociationProcess, self).setup()
self.inner_logger.debug(
'Setting up opPixelClassification and multicut_shell...')
# todo: replace opPixelClassification with catpy tile-getter
self.opPixelClassification, self.multicut_shell = setup_classifier_and_multicut(
*self.setup_args)
self.inner_logger.debug(
'opPixelClassification and multicut_shell set up')
Request.reset_thread_pool(1)
def test_lotsOfSmallRequests(self):
handlerCounter = [0]
handlerLock = threading.Lock()
def completionHandler( result, req ):
handlerLock.acquire()
handlerCounter[0] += 1
handlerLock.release()
requestCounter = [0]
requestLock = threading.Lock()
allRequests = []
# This closure randomly chooses to either (a) return immediately or (b) fire off more work
def someWork(depth, force=False, i=-1):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.5):
requests = []
for i in range(10):
req = Request(someWork, depth=depth-1, i=i)
req.notify(completionHandler, req=req)
requests.append(req)
allRequests.append(req)
requestLock.acquire()
requestCounter[0] += 1
requestLock.release()
for r in requests:
r.wait()
req = Request(someWork, depth=6, force=True)
def blubb(req):
pass
req.notify(blubb)
print "pausing graph"
global_thread_pool.pause()
global_thread_pool.unpause()
print "resumed graph"
req.wait()
print "request finished"
# Handler should have been called once for each request we fired
assert handlerCounter[0] == requestCounter[0]
print "finished testLotsOfSmallRequests"
for r in allRequests:
assert r.finished
print "waited for all subrequests"
def someWork(depth, force=False, i=0):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.5):
requests = []
for i in range(maxBreadth):
req = Request(someWork, depth=depth-1, i=i)
req.notify(writeToH5Py, index=(depth-1, i), req=req)
requests.append(req)
for r in requests:
r.wait()
def exportFinalSegmentation(self,
outputPath,
axisorder,
progressCallback=None):
assert self.FinalSegmentation.ready(
), "Can't export yet: The final segmentation isn't ready!"
logger.info("Starting Final Segmentation Export...")
opTranspose = OpReorderAxes(parent=self)
opTranspose.AxisOrder.setValue(axisorder)
opTranspose.Input.connect(self.FinalSegmentation)
f = h5py.File(outputPath, 'w')
opExporter = OpH5WriterBigDataset(parent=self)
opExporter.hdf5File.setValue(f)
opExporter.hdf5Path.setValue('split_result')
opExporter.Image.connect(opTranspose.Output)
if progressCallback is not None:
opExporter.progressSignal.subscribe(progressCallback)
req = Request(partial(self._runExporter, opExporter))
def cleanOps():
opExporter.cleanUp()
opTranspose.cleanUp()
def handleFailed(exc, exc_info):
cleanOps()
f.close()
import traceback
traceback.print_tb(exc_info[2])
msg = "Final Segmentation export FAILED due to the following error:\n{}".format(
exc)
logger.error(msg)
def handleFinished(result):
try:
cleanOps()
logger.info("FINISHED Final Segmentation Export")
finally:
f.close()
def handleCancelled():
cleanOps()
f.close()
logger.info("Final Segmentation export was cancelled!")
req.notify_failed(handleFailed)
req.notify_finished(handleFinished)
req.notify_cancelled(handleCancelled)
req.submit()
return req # Returned in case the user wants to cancel it.
def someWork(depth, force=False, i=0):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.5):
requests = []
for i in range(maxBreadth):
req = Request(someWork, depth=depth - 1, i=i)
req.notify(writeToH5Py, index=(depth - 1, i), req=req)
requests.append(req)
for r in requests:
r.wait()
def test_RequestLock():
assert Request.global_thread_pool.num_workers > 0, "This test must be used with the real threadpool."
lockA = RequestLock()
lockB = RequestLock()
def log_request_system_status():
status = (
"*************************\n"
+ "lockA.pending: {}\n".format(len(lockA._pendingRequests))
+ "lockB.pending: {}\n".format(len(lockB._pendingRequests))
# + "suspended Requests: {}\n".format( len(Request.global_suspend_set) )
+ "global job queue: {}\n".format(len(Request.global_thread_pool.unassigned_tasks))
)
for worker in Request.global_thread_pool.workers:
status += "{} queued tasks: {}\n".format(worker.name, len(worker.job_queue))
status += "*****************************************************"
logger.debug(status)
running = [True]
def periodic_status():
while running[0]:
time.sleep(0.5)
log_request_system_status()
# Uncomment these lines to print periodic status while the test runs...
status_thread = threading.Thread(target=periodic_status)
status_thread.daemon = True
status_thread.start()
try:
_impl_test_lock(lockA, lockB, Request, 1000)
except:
log_request_system_status()
running[0] = False
status_thread.join()
global paused
paused = False
Request.reset_thread_pool(Request.global_thread_pool.num_workers)
if lockA.locked():
lockA.release()
if lockB.locked():
lockB.release()
raise
log_request_system_status()
running[0] = False
status_thread.join()
def _update_rendering(self):
"""
Override from the base class.
"""
# This update has to be performed in a different thread to avoid a deadlock
# (Because this function is running in the context of a dirty notification!)
req = Request( self.__update_rendering )
def handle_rendering_failure( exc, exc_info ):
msg = "Exception raised during volume rendering update. See traceack above.\n"
log_exception( logger, msg, exc_info )
req.notify_failed( handle_rendering_failure )
req.submit()
def testBasic(self):
"""
Test the SimpleRequestCondition, which is like threading.Condition, but with a subset of the functionality.
(See the docs for details.)
"""
# num_workers = Request.global_thread_pool.num_workers
# Request.reset_thread_pool(num_workers=1)
N_ELEMENTS = 100
# It's tempting to simply use threading.Condition here,
# but that doesn't quite work if the thread calling wait() is also a worker thread.
# (threading.Condition uses threading.Lock() as it's 'waiter' lock, which blocks the entire worker.)
# cond = threading.Condition( RequestLock() )
cond = SimpleRequestCondition()
produced = []
consumed = []
def wait_for_all():
def f(i):
time.sleep(0.2 * random.random())
with cond:
produced.append(i)
cond.notify()
reqs = []
for i in range(N_ELEMENTS):
req = Request(partial(f, i))
reqs.append(req)
for req in reqs:
req.submit()
_consumed = consumed
with cond:
while len(_consumed) < N_ELEMENTS:
while len(_consumed) == len(produced):
cond.wait()
logger.debug("copying {} elements".format(len(produced) - len(consumed)))
_consumed += produced[len(_consumed) :]
# Force the request to run in a worker thread.
# This should catch failures that can occur if the Condition's "waiter" lock isn't a request lock.
req = Request(wait_for_all)
req.submit()
# Now block for completion
req.wait()
logger.debug("produced: {}".format(produced))
logger.debug("consumed: {}".format(consumed))
assert set(consumed) == set(range(N_ELEMENTS)), "Expected set(range(N_ELEMENTS)), got {}".format(consumed)
def _train_forests_with_feature_importance(forests,
X,
y,
feature_names,
export_path=None):
"""
Train all RFs (in parallel) and compute feature importances while doing so.
The importances table will be logged as INFO, and also exported to a file if export_path is given.
Returns: oobs and importances
"""
oobs = [None] * len(forests)
importances = [None] * len(forests)
def store_training_results(i, training_results):
oob, importance_results = training_results
oobs[i] = oob
importances[i] = importance_results
with Timer() as train_timer:
pool = RequestPool()
for i, forest in enumerate(forests):
req = Request(partial(forest.learnRFWithFeatureSelection, X,
y))
# save the training results
req.notify_finished(partial(store_training_results, i))
pool.add(req)
pool.wait()
logger.info("Training took, {} seconds".format(train_timer.seconds()))
# Forests may have different numbers of trees,
# so take a weighted average of their importances
tree_counts = [f.treeCount() for f in forests]
weights = numpy.array(tree_counts).astype(float)
weights /= weights.sum()
named_importances = collections.OrderedDict(
list(
zip(feature_names,
numpy.average(importances, weights=weights, axis=0))))
importance_table = generate_importance_table(named_importances,
sort="overall",
export_path=export_path)
logger.info(
"Feature importance measurements during training: \n{}".format(
importance_table))
return oobs, named_importances
def test_callWaitDuringCallback(self):
"""
When using request.notify(...) to handle request completions, the handler should be allowed to call request.wait().
Currently, this causes a hang somewhere in request.py.
"""
def handler(result, req):
return
req.wait()
def workFn():
pass
req = Request(workFn)
req.notify( handler, req=req )
req.wait()
def _handleCancelledRequest(self, roi):
# I can't think of a use-case for cancelling our child requests independent of our
assert Request.current_request_is_cancelled(), \
"You can cancel the parent request of this batch request action,"\
" but you can't cancel the child requests independently."
with self._condition:
self._condition.notify()
def someWork(depth, force=False, i=-1):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.8):
requests = []
for i in range(10):
req = Request(someWork, depth=depth - 1, i=i)
req.notify(completionHandler, req=req)
requests.append(req)
allRequests.append(req)
requestLock.acquire()
requestCounter[0] += 1
requestLock.release()
for r in requests:
r.wait()
def execute(self, slot, subindex, roi, result):
def compute_for_channel(output_channel, input_channel):
input_roi = numpy.array((roi.start, roi.stop))
input_roi[:, -1] = (input_channel, input_channel + 1)
input_req = self.Input(*input_roi)
# If possible, use the result array itself as a scratch area
if self.Input.meta.dtype == result.dtype:
input_req.writeInto(result[...,
output_channel:output_channel + 1])
input_data = input_req.wait()
input_data = input_data.astype(numpy.float32,
order='C',
copy=False)
input_data = input_data[..., 0] # drop channel axis
result[..., output_channel] = computeIntegralImage(input_data)
pool = RequestPool()
for output_channel, input_channel in enumerate(
range(roi.start[-1], roi.stop[-1])):
pool.add(
Request(
partial(compute_for_channel, output_channel,
input_channel)))
pool.wait()
def _handleCancelledRequest(self, roi):
# I can't think of a use-case for cancelling our child requests independent of our
assert Request.current_request_is_cancelled(), \
"You can cancel the parent request of this batch request action,"\
" but you can't cancel the child requests independently."
with self._condition:
self._condition.notify()
def _label(self, roi, result):
result = vigra.taggedView(result, axistags=self.Output.meta.axistags)
# get the background values
bg = self.Background[...].wait()
bg = vigra.taggedView(bg, axistags=self.Background.meta.axistags)
bg = bg.withAxes(*"ct")
assert np.all(
self.Background.meta.shape[0] == self.Input.meta.shape[0]
), "Shape of background values incompatible to shape of Input"
assert np.all(
self.Background.meta.shape[4] == self.Input.meta.shape[4]
), "Shape of background values incompatible to shape of Input"
# do labeling in parallel over channels and time slices
pool = RequestPool()
start = np.asarray(roi.start, dtype=np.int)
stop = np.asarray(roi.stop, dtype=np.int)
for ti, t in enumerate(range(roi.start[0], roi.stop[0])):
start[0], stop[0] = t, t + 1
for ci, c in enumerate(range(roi.start[4], roi.stop[4])):
start[4], stop[4] = c, c + 1
newRoi = SubRegion(self.Output, start=tuple(start), stop=tuple(stop))
resView = result[ti, ..., ci].withAxes(*"xyz")
req = Request(partial(self._label3d, newRoi, bg[c, t], resView))
pool.add(req)
logger.debug("{}: Computing connected components for ROI {} ...".format(self.name, roi))
pool.wait()
pool.clean()
logger.debug("{}: Connected components computed.".format(self.name))
def execute(self, slot, subindex, roi, result):
assert len(roi.start) == len(roi.stop) == len(self.Output.meta.shape)
assert slot == self.Output
t_ind = self.RawVolume.meta.axistags.index('t')
assert t_ind < len(self.RawVolume.meta.shape)
def compute_features_for_time_slice(res_t_ind, t):
axes4d = [
k for k in self.RawVolume.meta.getTaggedShape().keys()
if k in 'xyzc'
]
# Process entire spatial volume
s = [slice(None)] * len(self.RawVolume.meta.shape)
s[t_ind] = slice(t, t + 1)
s = tuple(s)
# Request in parallel
raw_req = self.RawVolume[s]
raw_req.submit()
label_req = self.LabelVolume[s]
label_req.submit()
if self.Atlas.ready():
atlasVolume = self.Atlas[s].wait()
atlasVolume = vigra.taggedView(
atlasVolume, axistags=self.Atlas.meta.axistags)
atlasVolume = atlasVolume.withAxes(*axes4d)
else:
atlasVolume = None
# Get results
rawVolume = raw_req.wait()
labelVolume = label_req.wait()
rawVolume = vigra.taggedView(rawVolume,
axistags=self.RawVolume.meta.axistags)
labelVolume = vigra.taggedView(
labelVolume, axistags=self.LabelVolume.meta.axistags)
# Convert to 4D (preserve axis order)
rawVolume = rawVolume.withAxes(*axes4d)
labelVolume = labelVolume.withAxes(*axes4d)
acc = self._extract(rawVolume, labelVolume, atlasVolume)
# Copy into the result
result[res_t_ind] = acc
# loop over requested time slices
pool = RequestPool()
for res_t_ind, t in enumerate(range(roi.start[t_ind],
roi.stop[t_ind])):
pool.add(
Request(partial(compute_features_for_time_slice, res_t_ind,
t)))
pool.wait()
return result
def addStack(self, roleIndex, laneIndex):
"""
The user clicked the "Import Stack Files" button.
"""
stackDlg = StackFileSelectionWidget(self)
stackDlg.exec_()
if stackDlg.result() != QDialog.Accepted :
return
files = stackDlg.selectedFiles
sequence_axis = stackDlg.sequence_axis
if len(files) == 0:
return
info = DatasetInfo()
info.filePath = os.path.pathsep.join( files )
prefix = os.path.commonprefix(files)
info.nickname = PathComponents(prefix).filenameBase
# Add an underscore for each wildcard digit
num_wildcards = len(files[-1]) - len(prefix) - len( os.path.splitext(files[-1])[1] )
info.nickname += "_"*num_wildcards
# Allow labels by default if this gui isn't being used for batch data.
info.allowLabels = ( self.guiMode == GuiMode.Normal )
info.fromstack = True
originalNumLanes = len(self.topLevelOperator.DatasetGroup)
if laneIndex is None or laneIndex == -1:
laneIndex = len(self.topLevelOperator.DatasetGroup)
if len(self.topLevelOperator.DatasetGroup) < laneIndex+1:
self.topLevelOperator.DatasetGroup.resize(laneIndex+1)
def importStack():
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested.emit()
# Serializer will update the operator for us, which will propagate to the GUI.
try:
self.serializer.importStackAsLocalDataset( info, sequence_axis )
try:
self.topLevelOperator.DatasetGroup[laneIndex][roleIndex].setValue(info)
except DatasetConstraintError as ex:
# Give the user a chance to repair the problem.
filename = files[0] + "\n...\n" + files[-1]
return_val = [False]
self.handleDatasetConstraintError( info, filename, ex, roleIndex, laneIndex, return_val )
if not return_val[0]:
# Not successfully repaired. Roll back the changes and give up.
self.topLevelOperator.DatasetGroup.resize(originalNumLanes)
finally:
self.parentApplet.busy = False
self.parentApplet.appletStateUpdateRequested.emit()
req = Request( importStack )
req.notify_finished( lambda result: self.showDataset(laneIndex, roleIndex) )
req.notify_failed( partial(self.handleFailedStackLoad, files, originalNumLanes ) )
req.submit()
def export(self, filename, hypothesesGraph, objectFeaturesSlot,
labelImageSlot, rawImageSlot):
"""Export the tracking solution stored in the hypotheses graph as a sequence of H5 files,
one per frame, containing the label image of that frame and which objects were part
of a move or a division.
:param filename: string of the FOLDER where to save the result
:param hypothesesGraph: hytra.core.hypothesesgraph.HypothesesGraph filled with a solution
:param objectFeaturesSlot: lazyflow.graph.InputSlot, connected to the RegionFeaturesAll output
of ilastik.applets.trackingFeatureExtraction.opTrackingFeatureExtraction.OpTrackingFeatureExtraction
:returns: True on success, False otherwise
"""
traxelIdPerTimestepToUniqueIdMap, uuidToTraxelMap = hypothesesGraph.getMappingsBetweenUUIDsAndTraxels(
)
timesteps = [t for t in traxelIdPerTimestepToUniqueIdMap.keys()]
result = hypothesesGraph.getSolutionDictionary()
mergers, detections, links, divisions = getMergersDetectionsLinksDivisions(
result, uuidToTraxelMap)
# group by timestep for event creation
mergersPerTimestep = getMergersPerTimestep(mergers, timesteps)
linksPerTimestep = getLinksPerTimestep(links, timesteps)
detectionsPerTimestep = getDetectionsPerTimestep(
detections, timesteps)
divisionsPerTimestep = getDivisionsPerTimestep(
divisions, linksPerTimestep, timesteps)
# save to disk in parallel
pool = RequestPool()
timeIndex = labelImageSlot.meta.axistags.index('t')
for timestep in traxelIdPerTimestepToUniqueIdMap.keys():
# extract current frame lable image
roi = [
slice(None) for i in range(len(labelImageSlot.meta.shape))
]
roi[timeIndex] = slice(int(timestep), int(timestep) + 1)
roi = tuple(roi)
labelImage = labelImageSlot[roi].wait()
if not os.path.exists(filename + '/H5-Event-Sequence'):
os.makedirs(filename + '/H5-Event-Sequence')
fn = os.path.join(
filename,
"H5-Event-Sequence/{0:05d}.h5".format(int(timestep)))
pool.add(
Request(
partial(writeEvents, int(timestep),
linksPerTimestep[timestep],
divisionsPerTimestep[timestep],
mergersPerTimestep[timestep],
detectionsPerTimestep[timestep], fn,
labelImage)))
pool.wait()
return True
def someWork(depth, force=False, i=-1):
#print 'depth=', depth, 'i=', i
if depth > 0 and (force or random.random() > 0.8):
requests = []
for i in range(10):
req = Request(someWork, depth=depth-1, i=i)
req.notify(completionHandler, req=req)
requests.append(req)
allRequests.append(req)
requestLock.acquire()
requestCounter[0] += 1
requestLock.release()
for r in requests:
r.wait()
def addStack(self, roleIndex, laneIndex):
"""
The user clicked the "Import Stack Files" button.
"""
stackDlg = StackFileSelectionWidget(self)
stackDlg.exec_()
if stackDlg.result() != QDialog.Accepted:
return
files = stackDlg.selectedFiles
sequence_axis = stackDlg.sequence_axis
if len(files) == 0:
return
cwd = self.topLevelOperator.WorkingDirectory.value
info = DatasetInfo(os.path.pathsep.join(files), cwd=cwd)
originalNumLanes = len(self.topLevelOperator.DatasetGroup)
if laneIndex is None or laneIndex == -1:
laneIndex = len(self.topLevelOperator.DatasetGroup)
if len(self.topLevelOperator.DatasetGroup) < laneIndex + 1:
self.topLevelOperator.DatasetGroup.resize(laneIndex + 1)
def importStack():
self.parentApplet.busy = True
self.parentApplet.appletStateUpdateRequested()
# Serializer will update the operator for us, which will propagate to the GUI.
try:
self.serializer.importStackAsLocalDataset(info, sequence_axis)
try:
self.topLevelOperator.DatasetGroup[laneIndex][
roleIndex].setValue(info)
except DatasetConstraintError as ex:
# Give the user a chance to repair the problem.
filename = files[0] + "\n...\n" + files[-1]
return_val = [False]
self.parentApplet.busy = False # required for possible fixing dialogs from DatasetConstraintError
self.parentApplet.appletStateUpdateRequested()
self.handleDatasetConstraintError(info, filename, ex,
roleIndex, laneIndex,
return_val)
if not return_val[0]:
# Not successfully repaired. Roll back the changes and give up.
self.topLevelOperator.DatasetGroup.resize(
originalNumLanes)
finally:
self.parentApplet.busy = False
self.parentApplet.appletStateUpdateRequested()
req = Request(importStack)
req.notify_finished(
lambda result: self.showDataset(laneIndex, roleIndex))
req.notify_failed(
partial(self.handleFailedStackLoad, files, originalNumLanes))
req.submit()
def execute(self, slot, subindex, roi, result):
"""
Simulate a cascade of requests, to make sure that the entire cascade is properly freed.
"""
roiShape = roi.stop - roi.start
def getResults1():
return numpy.indices(roiShape, self.Output.meta.dtype).sum()
def getResults2():
req = Request( getResults1 )
req.submit()
result[:] = req.wait()
return result
req = Request( getResults2 )
req.submit()
result[:] = req.wait()
return result
def execute_tasks(tasks):
"""
Executes the given list of tasks (functions) in the lazyflow threadpool.
"""
pool = RequestPool()
for task in tasks:
pool.add(Request(task))
pool.wait()
def replaceWithStack(self, roleIndex, laneIndex):
"""
The user clicked the "Import Stack Files" button.
"""
stackDlg = StackFileSelectionWidget(self)
stackDlg.exec_()
if stackDlg.result() != QDialog.Accepted :
return
files = stackDlg.selectedFiles
if len(files) == 0:
return
info = DatasetInfo()
info.filePath = "//".join( files )
prefix = os.path.commonprefix(files)
info.nickname = PathComponents(prefix).filenameBase + "..."
# Allow labels by default if this gui isn't being used for batch data.
info.allowLabels = ( self.guiMode == GuiMode.Normal )
info.fromstack = True
originalNumLanes = len(self.topLevelOperator.DatasetGroup)
if laneIndex is None:
laneIndex = self._findFirstEmptyLane(roleIndex)
if len(self.topLevelOperator.DatasetGroup) < laneIndex+1:
self.topLevelOperator.DatasetGroup.resize(laneIndex+1)
def importStack():
self.guiControlSignal.emit( ControlCommand.DisableAll )
# Serializer will update the operator for us, which will propagate to the GUI.
try:
self.serializer.importStackAsLocalDataset( info )
try:
self.topLevelOperator.DatasetGroup[laneIndex][roleIndex].setValue(info)
except DatasetConstraintError as ex:
# Give the user a chance to repair the problem.
filename = files[0] + "\n...\n" + files[-1]
if not self.handleDatasetConstraintError( info, filename, ex, roleIndex, laneIndex ):
self.topLevelOperator.DatasetGroup.resize(originalNumLanes)
finally:
self.guiControlSignal.emit( ControlCommand.Pop )
req = Request( importStack )
req.notify_failed( partial(self.handleFailedStackLoad, files, originalNumLanes ) )
req.submit()
def _resolveMergers(self, hypothesesGraph, model):
'''
run merger resolution on the hypotheses graph which contains the current solution
'''
logger.info("Resolving mergers.")
parameters = self.Parameters.value
withTracklets = parameters['withTracklets']
originalGraph = hypothesesGraph.referenceTraxelGraph if withTracklets else hypothesesGraph
resolvedMergersDict = {}
# Enable full graph computation for animal tracking workflow
withFullGraph = False
if 'withAnimalTracking' in parameters and parameters['withAnimalTracking']: # TODO: Setting this parameter outside of the track() function (on AnimalConservationTrackingWorkflow) is not desirable
withFullGraph = True
logger.info("Computing full graph on merger resolver (Only enabled on animal tracking workflow)")
mergerResolver = IlastikMergerResolver(originalGraph, pluginPaths=self.pluginPaths, withFullGraph=withFullGraph)
# Check if graph contains mergers, otherwise skip merger resolving
if not mergerResolver.mergerNum:
logger.info("Graph contains no mergers. Skipping merger resolving.")
else:
# Fit and refine merger nodes using a GMM
# It has to be done per time-step in order to aviod loading the whole video on RAM
traxelIdPerTimestepToUniqueIdMap, uuidToTraxelMap = getMappingsBetweenUUIDsAndTraxels(model)
timesteps = [int(t) for t in traxelIdPerTimestepToUniqueIdMap.keys()]
timesteps.sort()
timeIndex = self.LabelImage.meta.axistags.index('t')
for timestep in timesteps:
roi = [slice(None) for i in range(len(self.LabelImage.meta.shape))]
roi[timeIndex] = slice(timestep, timestep+1)
roi = tuple(roi)
labelImage = self.LabelImage[roi].wait()
# Get coordinates for object IDs in label image. Used by GMM merger fit.
objectIds = vigra.analysis.unique(labelImage[0,...,0])
maxObjectId = max(objectIds)
coordinatesForIds = {}
pool = RequestPool()
for objectId in objectIds:
pool.add(Request(partial(mergerResolver.getCoordinatesForObjectId, coordinatesForIds, labelImage[0, ..., 0], timestep, objectId)))
# Run requests to get object ID coordinates
pool.wait()
# Fit mergers and store fit info in nodes
if coordinatesForIds:
mergerResolver.fitAndRefineNodesForTimestep(coordinatesForIds, maxObjectId, timestep)
# Compute object features, re-run flow solver, update model and result, and get merger dictionary
resolvedMergersDict = mergerResolver.run()
return resolvedMergersDict
def read(self, view_roi, result_out):
"""
roi: (start, stop) tuples, ordered according to description.output_axes
roi should be relative to the view
"""
output_axes = self.description.output_axes
roi_transposed = list(zip(*view_roi))
roi_dict = dict(list(zip(output_axes, roi_transposed)))
view_roi = list(zip(*(roi_dict["z"], roi_dict["y"], roi_dict["x"])))
# First, normalize roi and result to zyx order
result_out = vigra.taggedView(result_out, output_axes)
result_out = result_out.withAxes(*"zyx")
assert numpy.array(view_roi).shape == (2, 3), "Invalid roi for 3D volume: {}".format(view_roi)
view_roi = numpy.array(view_roi)
assert (result_out.shape == (view_roi[1] - view_roi[0])).all()
# User gave roi according to the view output.
# Now offset it find global roi.
roi = view_roi + self.description.view_origin_zyx
tile_blockshape = (1,) + tuple(self.description.tile_shape_2d_yx)
tile_starts = getIntersectingBlocks(tile_blockshape, roi)
pool = RequestPool()
for tile_start in tile_starts:
tile_roi_in = getBlockBounds(self.description.bounds_zyx, tile_blockshape, tile_start)
tile_roi_in = numpy.array(tile_roi_in)
# This tile's portion of the roi
intersecting_roi = getIntersection(roi, tile_roi_in)
intersecting_roi = numpy.array(intersecting_roi)
# Compute slicing within destination array and slicing within this tile
destination_relative_intersection = numpy.subtract(intersecting_roi, roi[0])
tile_relative_intersection = intersecting_roi - tile_roi_in[0]
# Get a view to the output slice
result_region = result_out[roiToSlice(*destination_relative_intersection)]
rest_args = self._get_rest_args(tile_blockshape, tile_roi_in)
if self.description.tile_url_format.startswith("http"):
retrieval_fn = partial(self._retrieve_remote_tile, rest_args, tile_relative_intersection, result_region)
else:
retrieval_fn = partial(self._retrieve_local_tile, rest_args, tile_relative_intersection, result_region)
PARALLEL_REQ = True
if PARALLEL_REQ:
pool.add(Request(retrieval_fn))
else:
# execute serially (leave the pool empty)
retrieval_fn()
if PARALLEL_REQ:
with Timer() as timer:
pool.wait()
logger.info("Loading {} tiles took a total of {}".format(len(tile_starts), timer.seconds()))
def test_basic(self):
def someWork():
time.sleep(0.001)
#print "producer finished"
def callback(s):
pass
def test(s):
req = Request(someWork)
req.notify(callback)
req.wait()
time.sleep(0.001)
print s
return s
req = Request(test, s="hallo !")
req.notify(callback)
assert req.wait() == "hallo !"
requests = []
for i in range(10):
req = Request(test, s="hallo %d" % i)
requests.append(req)
for r in requests:
r.wait()
def main():
parser = argparse.ArgumentParser()
parser.add_argument('grayscale', help='example: my-grayscale.h5/volume')
parser.add_argument('classifier', help='example: my-file.h5/forest')
parser.add_argument('filter_specs', help='json file containing filter list')
parser.add_argument('output_path', help='example: my-predictions.h5/volume')
parser.add_argument('--compute-blockwise', help='Compute blockwise instead of as a whole', action='store_true')
parser.add_argument('--thread-count', help='The threadpool size', default=0, type=int)
args = parser.parse_args()
# Show log messages on the console.
logger.setLevel(logging.INFO)
logger.addHandler( logging.StreamHandler(sys.stdout) )
Request.reset_thread_pool(args.thread_count)
load_and_predict( args.grayscale, args.classifier, args.filter_specs, args.output_path, args.compute_blockwise )
logger.info("DONE.")
def _train_forests(forests, X, y):
"""
Train all RFs (in parallel), and return the oobs.
"""
oobs = [None] * len(forests)
def store_oob_results(i, oob):
oobs[i] = oob
with Timer() as train_timer:
pool = RequestPool()
for i, forest in enumerate(forests):
req = Request( partial(forest.learnRF, X, y) )
# save the oob results
req.notify_finished( partial( store_oob_results, i ) )
pool.add( req )
pool.wait()
logger.info("Training took, {} seconds".format( train_timer.seconds() ) )
return oobs
def compute_all_features():
# Compute features in parallel
pool = RequestPool()
for t in range(tMax):
pool.add(
Request(
partial(compute_features_for_frame, tIndex, t,
features)))
pool.wait()
def get(self, roi):
"""This method is used to retrieve the actual content of a Slot.
:param roi: the region of interest, e.g. a subregion in the
case of an ArrayLike stype
:param destination: this may define a destination area for the
request, for example a ndarray into which the results should
be written in the case of an ArrayLike stype
Returns:
a request.Request object.
"""
if self._value is not None:
# this handles the case of an inputslot
# having a ._value
# --> construct cheaper request object for this case
result = self.stype.writeIntoDestination(None, self._value, roi)
return ValueRequest(result)
elif self.partner is not None:
# this handles the case of an inputslot
# --> just relay the request
return self.partner.get(roi)
else:
if not self.ready():
msg = "Can't get data from slot {}.{} yet."\
" It isn't ready."\
"First upstream problem slot is: {}"
msg = msg.format( self.getRealOperator().__class__, self.name, Slot._findUpstreamProblemSlot(self) )
assert self.ready(), msg
# If someone is asking for data from an inputslot that has
# no value and no partner, then something is wrong.
assert self._type != "input", "This inputSlot has no value and no partner. You can't ask for its data yet!"
# normal (outputslot) case
# --> construct heavy request object..
execWrapper = Slot.RequestExecutionWrapper(self, roi)
request = Request(execWrapper)
# We must decrement the execution count even if the
# request is cancelled
request.notify_cancelled(execWrapper.handleCancel)
return request
def exportFinalSegmentation(self, outputPath, axisorder, progressCallback=None):
assert self.FinalSegmentation.ready(), "Can't export yet: The final segmentation isn't ready!"
logger.info("Starting Final Segmentation Export...")
opTranspose = OpReorderAxes( parent=self )
opTranspose.AxisOrder.setValue( axisorder )
opTranspose.Input.connect( self.FinalSegmentation )
f = h5py.File(outputPath, 'w')
opExporter = OpH5WriterBigDataset(parent=self)
opExporter.hdf5File.setValue( f )
opExporter.hdf5Path.setValue( 'split_result' )
opExporter.Image.connect( opTranspose.Output )
if progressCallback is not None:
opExporter.progressSignal.subscribe( progressCallback )
req = Request( partial(self._runExporter, opExporter) )
def cleanOps():
opExporter.cleanUp()
opTranspose.cleanUp()
def handleFailed( exc, exc_info ):
cleanOps()
f.close()
import traceback
traceback.print_tb(exc_info[2])
msg = "Final Segmentation export FAILED due to the following error:\n{}".format( exc )
logger.error( msg )
def handleFinished( result ):
try:
cleanOps()
logger.info("FINISHED Final Segmentation Export")
finally:
f.close()
def handleCancelled():
cleanOps()
f.close()
logger.info( "Final Segmentation export was cancelled!" )
req.notify_failed( handleFailed )
req.notify_finished( handleFinished )
req.notify_cancelled( handleCancelled )
req.submit()
return req # Returned in case the user wants to cancel it.
def _train_forests_with_feature_importance(forests, X, y, feature_names, export_path=None):
"""
Train all RFs (in parallel) and compute feature importances while doing so.
The importances table will be logged as INFO, and also exported to a file if export_path is given.
Returns: oobs and importances
"""
oobs = [None] * len(forests)
importances = [None] * len(forests)
def store_training_results(i, training_results):
oob, importance_results = training_results
oobs[i] = oob
importances[i] = importance_results
with Timer() as train_timer:
pool = RequestPool()
for i, forest in enumerate(forests):
req = Request(partial(forest.learnRFWithFeatureSelection, X, y))
# save the training results
req.notify_finished(partial(store_training_results, i))
pool.add(req)
pool.wait()
logger.info("Training took, {} seconds".format(train_timer.seconds()))
# Forests may have different numbers of trees,
# so take a weighted average of their importances
tree_counts = [f.treeCount() for f in forests]
weights = numpy.array(tree_counts).astype(float)
weights /= weights.sum()
named_importances = collections.OrderedDict(
list(zip(feature_names, numpy.average(importances, weights=weights, axis=0)))
)
importance_table = generate_importance_table(named_importances, sort="overall", export_path=export_path)
logger.info("Feature importance measurements during training: \n{}".format(importance_table))
return oobs, named_importances
def importStackFromGlobString(self, globString):
"""
The word 'glob' is used loosely here. See the OpStackLoader operator for details.
"""
globString = globString.replace("\\","/")
info = DatasetInfo()
info.filePath = globString
# Allow labels by default if this gui isn't being used for batch data.
info.allowLabels = ( self.guiMode == GuiMode.Normal )
def importStack():
self.guiControlSignal.emit( ControlCommand.DisableAll )
# Serializer will update the operator for us, which will propagate to the GUI.
try:
self.serializer.importStackAsLocalDataset( info )
finally:
self.guiControlSignal.emit( ControlCommand.Pop )
req = Request( importStack )
req.notify_failed( partial(self.handleFailedStackLoad, globString ) )
req.submit()
def wait_for_all():
def f(i):
time.sleep(0.2*random.random())
with cond:
produced.append(i)
cond.notify()
reqs = []
for i in range(N_ELEMENTS):
req = Request( partial(f, i) )
reqs.append( req )
for req in reqs:
req.submit()
_consumed = consumed
with cond:
while len(_consumed) < N_ELEMENTS:
while len(_consumed) == len(produced):
cond.wait()
logger.debug( "copying {} elements".format( len(produced) - len(consumed) ) )
_consumed += produced[len(_consumed):]