def _augmentFeatureNames(self, features):
# Take a dictionary of feature names, augment it by default features and set to Features() slot
feature_names = features
feature_names_with_default = deepcopy(feature_names)
#expand the feature list by our default features
logger.debug(
"attaching default features {} to vigra features {}".format(
default_features, feature_names))
plugin = pluginManager.getPluginByName("Standard Object Features",
"ObjectFeatures")
all_default_props = plugin.plugin_object.fill_properties(
default_features) #fill in display name and such
feature_names_with_default[default_features_key] = all_default_props
if not "Standard Object Features" in list(feature_names.keys()):
# The user has not selected any standard features. Add them now
feature_names_with_default["Standard Object Features"] = {}
for default_feature_name, default_feature_props in default_features.items(
):
if default_feature_name not in feature_names_with_default[
"Standard Object Features"]:
# this feature has not been selected by the user, add it now.
feature_names_with_default["Standard Object Features"][
default_feature_name] = all_default_props[
default_feature_name]
feature_names_with_default["Standard Object Features"][
default_feature_name]["selected"] = False
return feature_names_with_default
def _export_with_plugin(self, lane_index: int, checkOverwriteFiles: bool, pluginName: str) -> bool:
argsSlot = self.topLevelOperator.AdditionalPluginArguments
pluginInfo = pluginManager.getPluginByName(pluginName, category="TrackingExportFormats")
if pluginInfo is None:
logger.error("Could not find selected plugin %s", pluginName)
return False
plugin = pluginInfo.plugin_object
logger.info("Exporting tracking result using %s", pluginName)
name_format = self.topLevelOperator.getLane(lane_index).OutputFilenameFormat.value
filename = self.getPartiallyFormattedName(lane_index, name_format)
if plugin.exportsToFile and not os.path.basename(filename):
filename = os.path.join(filename, 'pluginExport.txt')
if not filename:
logger.error("Cannot export from plugin with empty output filename")
return False
self.progressSignal(-1)
try:
status = self._pluginExportFunc(lane_index, filename, plugin, checkOverwriteFiles, argsSlot)
finally:
self.progressSignal(100)
if not status:
return False
logger.info("Export done")
return True
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# `checkOverwriteFiles` parameter ensures we check only once for files that could be overwritten, pop up
# the MessageBox and then don't export. For the next round we click the export button,
# we really want it to export, so checkOverwriteFiles=False.
# Plugin export if selected
logger.info(
"Export source is: " +
self.dataExportApplet.topLevelOperator.SelectedExportSource.value)
if self.dataExportApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportApplet.topLevelOperator.SelectedPlugin.value
additionalPluginArgumentsSlot = self.dataExportApplet.topLevelOperator.AdditionalPluginArguments
exportPluginInfo = pluginManager.getPluginByName(
selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" %
exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" %
selectedPlugin)
name_format = self.dataExportApplet.topLevelOperator.getLane(
lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(
lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = partially_formatted_name
if filename is None or len(str(filename)) == 0:
logger.error(
"Cannot export from plugin with empty output filename")
return True
self.dataExportApplet.progressSignal(-1)
exportStatus = self.trackingApplet.topLevelOperator.getLane(
lane_index).exportPlugin(filename, exportPlugin,
checkOverwriteFiles,
additionalPluginArgumentsSlot)
self.dataExportApplet.progressSignal(100)
if not exportStatus:
return False
logger.info("Export done")
return True
return True
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# Plugin export if selected
logger.info("Export source is: " + self.dataExportTrackingApplet.
topLevelOperator.SelectedExportSource.value)
print("in post_process_lane_export")
if self.dataExportTrackingApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportTrackingApplet.topLevelOperator.SelectedPlugin.value
exportPluginInfo = pluginManager.getPluginByName(
selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" %
exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" %
selectedPlugin)
name_format = self.dataExportTrackingApplet.topLevelOperator.getLane(
lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(
lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = partially_formatted_name
if filename is None or len(str(filename)) == 0:
logger.error(
"Cannot export from plugin with empty output filename")
return True
self.dataExportTrackingApplet.progressSignal(-1)
exportStatus = self.trackingApplet.topLevelOperator.getLane(
lane_index).exportPlugin(filename, exportPlugin,
checkOverwriteFiles)
self.dataExportTrackingApplet.progressSignal(100)
if not exportStatus:
return False
logger.info("Export done")
return True
return True
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# `checkOverwriteFiles` parameter ensures we check only once for files that could be overwritten, pop up
# the MessageBox and then don't export. For the next round we click the export button,
# we really want it to export, so checkOverwriteFiles=False.
# Plugin export if selected
logger.info("Export source is: " + self.dataExportApplet.topLevelOperator.SelectedExportSource.value)
if self.dataExportApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportApplet.topLevelOperator.SelectedPlugin.value
exportPluginInfo = pluginManager.getPluginByName(selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" % exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" % selectedPlugin)
name_format = self.dataExportApplet.topLevelOperator.getLane(lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = partially_formatted_name
if filename is None or len(str(filename)) == 0:
logger.error("Cannot export from plugin with empty output filename")
return True
self.dataExportApplet.progressSignal(-1)
exportStatus = self.trackingApplet.topLevelOperator.getLane(lane_index).exportPlugin(filename, exportPlugin, checkOverwriteFiles)
self.dataExportApplet.progressSignal(100)
if not exportStatus:
return False
logger.info("Export done")
return True
return True
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# Plugin export if selected
logger.info("Export source is: " + self.dataExportTrackingApplet.topLevelOperator.SelectedExportSource.value)
print("in post_process_lane_export")
if self.dataExportTrackingApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportTrackingApplet.topLevelOperator.SelectedPlugin.value
exportPluginInfo = pluginManager.getPluginByName(selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" % exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" % selectedPlugin)
name_format = self.dataExportTrackingApplet.topLevelOperator.getLane(lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = partially_formatted_name
if filename is None or len(str(filename)) == 0:
logger.error("Cannot export from plugin with empty output filename")
return True
self.dataExportTrackingApplet.progressSignal(-1)
exportStatus = self.trackingApplet.topLevelOperator.getLane(lane_index).exportPlugin(filename, exportPlugin, checkOverwriteFiles)
self.dataExportTrackingApplet.progressSignal(100)
if not exportStatus:
return False
logger.info("Export done")
return True
return True
def _augmentFeatureNames(self, features):
# Take a dictionary of feature names, augment it by default features and set to Features() slot
feature_names = features
feature_names_with_default = deepcopy(feature_names)
#expand the feature list by our default features
logger.debug("attaching default features {} to vigra features {}".format(default_features, feature_names))
plugin = pluginManager.getPluginByName("Standard Object Features", "ObjectFeatures")
all_default_props = plugin.plugin_object.fill_properties(default_features) #fill in display name and such
feature_names_with_default[default_features_key] = all_default_props
if not "Standard Object Features" in list(feature_names.keys()):
# The user has not selected any standard features. Add them now
feature_names_with_default["Standard Object Features"] = {}
for default_feature_name, default_feature_props in default_features.items():
if default_feature_name not in feature_names_with_default["Standard Object Features"]:
# this feature has not been selected by the user, add it now.
feature_names_with_default["Standard Object Features"][default_feature_name] = all_default_props[default_feature_name]
feature_names_with_default["Standard Object Features"][default_feature_name]["selected"] = False
return feature_names_with_default
def _extract(self, image, labels):
if not (image.ndim == labels.ndim == 4):
raise Exception("both images must be 4D. raw image shape: {}"
" label image shape: {}".format(image.shape, labels.shape))
# FIXME: maybe simplify? taggedShape should be easier here
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
logger.debug("Computing default features")
feature_names = deepcopy(self.Features([]).wait())
# do global features
logger.debug("computing global features")
extra_features_computed = False
global_features = {}
selected_vigra_features = []
for plugin_name, feature_dict in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
if plugin_name == "Standard Object Features":
#expand the feature list by our default features
logger.debug("attaching default features {} to vigra features {}".format(default_features, feature_dict))
selected_vigra_features = feature_dict.keys()
feature_dict.update(default_features)
extra_features_computed = True
global_features[plugin_name] = plugin.plugin_object.compute_global(image, labels, feature_dict, axes)
extrafeats = {}
if extra_features_computed:
for feat_key in default_features:
feature = None
if feat_key in selected_vigra_features:
#we wanted that feature independently
feature = global_features["Standard Object Features"][feat_key]
else:
feature = global_features["Standard Object Features"].pop(feat_key)
feature_names["Standard Object Features"].pop(feat_key)
extrafeats[feat_key] = feature
else:
logger.debug("default features not computed, computing separately")
extrafeats_acc = vigra.analysis.extractRegionFeatures(image[slc3d].squeeze().astype(np.float32), labels.squeeze(),
default_features.keys(),
ignoreLabel=0)
#remove the 0th object, we'll add it again later
for k, v in extrafeats_acc.iteritems():
extrafeats[k]=v[1:]
if len(v.shape)==1:
extrafeats[k]=extrafeats[k].reshape(extrafeats[k].shape+(1,))
extrafeats = dict((k.replace(' ', ''), v)
for k, v in extrafeats.iteritems())
mincoords = extrafeats["Coord<Minimum>"]
maxcoords = extrafeats["Coord<Maximum>"]
nobj = mincoords.shape[0]
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = defaultdict(lambda: defaultdict(list))
margin = max_margin(feature_names)
has_local_features = {}
for plugin_name, feature_dict in feature_names.iteritems():
has_local_features[plugin_name] = False
for features in feature_dict.itervalues():
if 'margin' in features:
has_local_features[plugin_name] = True
break
if np.any(margin) > 0:
#starting from 0, we stripped 0th background object in global computation
for i in range(0, nobj):
logger.debug("processing object {}".format(i))
extent = self.compute_extent(i, image, mincoords, maxcoords, axes, margin)
rawbbox = self.compute_rawbbox(image, extent, axes)
#it's i+1 here, because the background has label 0
binary_bbox = np.where(labels[tuple(extent)] == i+1, 1, 0).astype(np.bool)
for plugin_name, feature_dict in feature_names.iteritems():
if not has_local_features[plugin_name]:
continue
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(rawbbox, binary_bbox, feature_dict, axes)
local_features[plugin_name] = dictextend(local_features[plugin_name], feats)
logger.debug("computing done, removing failures")
# remove local features that failed
for pname, pfeats in local_features.iteritems():
for key in pfeats.keys():
value = pfeats[key]
try:
pfeats[key] = np.vstack(list(v.reshape(1, -1) for v in value))
except:
logger.warn('feature {} failed'.format(key))
del pfeats[key]
# merge the global and local features
logger.debug("removed failed, merging")
all_features = {}
plugin_names = set(global_features.keys()) | set(local_features.keys())
for name in plugin_names:
d1 = global_features.get(name, {})
d2 = local_features.get(name, {})
all_features[name] = dict(d1.items() + d2.items())
all_features[default_features_key]=extrafeats
# reshape all features
for pfeats in all_features.itervalues():
for key, value in pfeats.iteritems():
if value.shape[0] != nobj:
raise Exception('feature {} does not have enough rows, {} instead of {}'.format(key, value.shape[0], nobj))
# because object classification operator expects nobj to
# include background. FIXME: we should change that assumption.
value = np.vstack((np.zeros(value.shape[1]),
value))
value = value.astype(np.float32) #turn Nones into numpy.NaNs
assert value.dtype == np.float32
assert value.shape[0] == nobj+1
assert value.ndim == 2
pfeats[key] = value
logger.debug("merged, returning")
return all_features
def compute_for_one_plugin(plugin_name, feature_dict):
plugin_inner = pluginManager.getPluginByName(
plugin_name, "ObjectFeatures")
global_features[
plugin_name] = plugin_inner.plugin_object.compute_global(
image, labels, feature_dict, axes)
def _extract(self, image, labels):
if not (image.ndim == labels.ndim == 4):
raise Exception("both images must be 4D. raw image shape: {}"
" label image shape: {}".format(image.shape, labels.shape))
# FIXME: maybe simplify?
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
image = np.asarray(image, dtype=np.float32)
labels = np.asarray(labels, dtype=np.uint32)
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
logger.debug("Computing default features")
#FIXME: clamp the global vigra features here
extrafeats = vigra.analysis.extractRegionFeatures(image[slc3d], labels,
default_features,
ignoreLabel=0)
logger.debug("computed default features")
extrafeats = dict((k.replace(' ', ''), v)
for k, v in extrafeats.iteritems())
mincoords = extrafeats["Coord<Minimum>"]
maxcoords = extrafeats["Coord<Maximum>"]
nobj = mincoords.shape[0]
feature_names = self.Features([]).wait()
# do global features
global_features = {}
for plugin_name, feature_list in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
global_features[plugin_name] = plugin.plugin_object.compute_global(image, labels, feature_list, axes)
logger.debug("computing global features")
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = defaultdict(lambda: defaultdict(list))
margin = max_margin(feature_names)
if np.any(margin) > 0:
for i in range(1, nobj):
logger.debug("processing object {}".format(i))
extent = self.compute_extent(i, image, mincoords, maxcoords, axes, margin)
rawbbox = self.compute_rawbbox(image, extent, axes)
binary_bbox = np.where(labels[tuple(extent)] == i, 1, 0).astype(np.bool)
for plugin_name, feature_list in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(rawbbox, binary_bbox, feature_list, axes)
local_features[plugin_name] = dictextend(local_features[plugin_name], feats)
logger.debug("computing done, removing failures")
# remove local features that failed
for pname, pfeats in local_features.iteritems():
for key in pfeats.keys():
value = pfeats[key]
try:
pfeats[key] = np.vstack(list(v.reshape(1, -1) for v in value))
except:
logger.warn('feature {} failed'.format(key))
del pfeats[key]
# merge the global and local features
logger.debug("removed failed, merging")
all_features = {}
plugin_names = set(global_features.keys()) | set(local_features.keys())
for name in plugin_names:
d1 = global_features.get(name, {})
d2 = local_features.get(name, {})
all_features[name] = dict(d1.items() + d2.items())
# reshape all features
for pfeats in all_features.itervalues():
for key, value in pfeats.iteritems():
if value.shape[0] != nobj - 1:
raise Exception('feature {} does not have enough rows')
# because object classification operator expects nobj to
# include background. we should change that assumption.
value = np.vstack((np.zeros(value.shape[1]),
value))
value = value.astype(np.float32) #turn Nones into numpy.NaNs
assert value.dtype == np.float32
assert value.shape[0] == nobj
assert value.ndim == 2
pfeats[key] = value
logger.debug("merged, returning")
# add features needed by downstream applets. these should be
# removed before classification.
all_features[default_features_key] = extrafeats
return all_features
def populate(self):
#self.ui.treeWidget.setColumnCount(2)
for pluginName, features in self.featureDict.items():
if pluginName=="TestFeatures" and not ilastik_config.getboolean("ilastik", "debug"):
continue
parent = QTreeWidgetItem(self.ui.treeWidget)
parent.setText(0, pluginName)
parent.setFlags(Qt.ItemIsUserCheckable | Qt.ItemIsEnabled)
# hack to ensure checkboxes visible
parent.setCheckState(0, Qt.Checked)
parent.setCheckState(0, Qt.Unchecked)
parent.setExpanded(False)
self.countChecked[pluginName]=0
self.countAll[pluginName]=len(self.featureDict[pluginName])
advanced_names = []
simple_names = []
selected_names = []
groups = set()
plugin = pluginManager.getPluginByName(pluginName, "ObjectFeatures")
features_with_props = deepcopy(features)
if plugin is not None:
plugin.plugin_object.fill_properties(features_with_props)
for name in sorted(features.keys()):
parameters = features[name]
for prop, prop_value in features_with_props[name].items():
if not prop in list(parameters.keys()):
# this property has not been added yet (perhaps the feature dictionary has been read from file)
# set it now
parameters[prop] = prop_value
try:
if parameters['advanced'] is True:
advanced_names.append(name)
else:
simple_names.append(name)
except KeyError:
simple_names.append(name)
try:
groups.add(parameters["group"])
except KeyError:
pass
if pluginName in self.selectedFeatures:
if name in self.selectedFeatures[pluginName]:
selected_names.append(name)
gr_items = {}
for gr in groups:
gr_items[gr] = QTreeWidgetItem(parent)
gr_items[gr].setText(0, gr)
#gr_items[gr].setFlags(Qt.ItemIsEnabled | Qt.ItemIsUserCheckable)
gr_items[gr].setExpanded(True)
for name in simple_names+advanced_names:
if name in advanced_names and (not name in selected_names):
# do not display advanced features, if they have not been selected previously
continue
parameters = features[name]
if "group" in parameters:
item = QTreeWidgetItem(gr_items[parameters["group"]])
item.group_name = parameters["group"]
else:
item = QTreeWidgetItem(parent)
if 'displaytext' in parameters:
itemtext = parameters['displaytext']
else:
itemtext = name
item.setText(0, itemtext)
item.feature_id = name
item.setFlags(Qt.ItemIsUserCheckable | Qt.ItemIsEnabled)
if 'tooltip' in parameters:
item.setToolTip(0, parameters['tooltip'])
# hack to ensure checkboxes visible
item.setCheckState(0, Qt.Checked)
item.setCheckState(0, Qt.Unchecked)
if name in selected_names:
item.setCheckState(0, Qt.Checked)
self.countChecked[pluginName]+=1
if self.countChecked[pluginName] == 0:
parent.setCheckState(0, Qt.Unchecked)
elif self.countChecked[pluginName] == self.countAll[pluginName]:
parent.setCheckState(0, Qt.Checked)
else:
parent.setCheckState(0, Qt.PartiallyChecked)
self.updateToolTip(parent)
# facilitates switching of the CheckBox when clicking on the Text of a QTreeWidgetItem
self.ui.treeWidget.setCurrentItem(None)
def _extract(self, image, labels, atlas=None):
if not (image.ndim == labels.ndim == 4):
raise Exception("both images must be 4D. raw image shape: {}"
" label image shape: {}".format(
image.shape, labels.shape))
# FIXME: maybe simplify? taggedShape should be easier here
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
#These are the feature names, selected by the user and the default feature names.
feature_names = deepcopy(self.Features([]).wait())
feature_names = self._augmentFeatureNames(feature_names)
# do global features
logger.debug("Computing global and default features")
global_features = {}
pool = RequestPool()
def compute_for_one_plugin(plugin_name, feature_dict):
plugin_inner = pluginManager.getPluginByName(
plugin_name, "ObjectFeatures")
global_features[
plugin_name] = plugin_inner.plugin_object.compute_global(
image, labels, feature_dict, axes)
for plugin_name, feature_dict in feature_names.items():
if plugin_name != default_features_key:
pool.add(
Request(
partial(compute_for_one_plugin, plugin_name,
feature_dict)))
pool.wait()
extrafeats = {}
for feat_key in default_features:
try:
sel = feature_names["Standard Object Features"][feat_key][
"selected"]
except KeyError:
# we don't always set this property to True, sometimes it's just not there. The only important
# thing is that it's not False
sel = True
if not sel:
# This feature has not been selected by the user. Remove it from the computed dict into a special dict
# for default features
feature = global_features["Standard Object Features"].pop(
feat_key)
else:
feature = global_features["Standard Object Features"][feat_key]
extrafeats[feat_key] = feature
if atlas is not None:
extrafeats['AtlasMapping'] = self._createAtlasMapping(
extrafeats['RegionCenter'], atlas)
extrafeats = dict(
(k.replace(' ', ''), v) for k, v in extrafeats.items())
mincoords = extrafeats["Coord<Minimum>"].astype(int)
maxcoords = extrafeats["Coord<Maximum>"].astype(int)
nobj = mincoords.shape[0]
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = collections.defaultdict(
lambda: collections.defaultdict(list))
margin = max_margin(feature_names)
has_local_features = {}
for plugin_name, feature_dict in feature_names.items():
has_local_features[plugin_name] = False
for features in feature_dict.values():
if 'margin' in features:
has_local_features[plugin_name] = True
break
if numpy.any(margin) > 0:
#starting from 0, we stripped 0th background object in global computation
for i in range(0, nobj):
logger.debug("processing object {}".format(i))
extent = self.compute_extent(i, image, mincoords, maxcoords,
axes, margin)
rawbbox = self.compute_rawbbox(image, extent, axes)
#it's i+1 here, because the background has label 0
binary_bbox = numpy.where(labels[tuple(extent)] == i + 1, 1,
0).astype(numpy.bool)
for plugin_name, feature_dict in feature_names.items():
if not has_local_features[plugin_name]:
continue
plugin = pluginManager.getPluginByName(
plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(
rawbbox, binary_bbox, feature_dict, axes)
local_features[plugin_name] = dictextend(
local_features[plugin_name], feats)
logger.debug("computing done, removing failures")
# remove local features that failed
for pname, pfeats in local_features.items():
for key in list(pfeats.keys()):
value = pfeats[key]
try:
pfeats[key] = numpy.vstack(
list(v.reshape(1, -1) for v in value))
except:
logger.warning('feature {} failed'.format(key))
del pfeats[key]
# merge the global and local features
logger.debug("removed failed, merging")
all_features = {}
plugin_names = set(global_features.keys()) | set(local_features.keys())
for name in plugin_names:
d1 = global_features.get(name, {})
d2 = local_features.get(name, {})
all_features[name] = dict(list(d1.items()) + list(d2.items()))
all_features[default_features_key] = extrafeats
# reshape all features
for pfeats in all_features.values():
for key, value in pfeats.items():
if value.shape[0] != nobj:
raise Exception(
'feature {} does not have enough rows, {} instead of {}'
.format(key, value.shape[0], nobj))
# because object classification operator expects nobj to
# include background. FIXME: we should change that assumption.
value = numpy.vstack((numpy.zeros(value.shape[1]), value))
value = value.astype(
numpy.float32) #turn Nones into numpy.NaNs
assert value.dtype == numpy.float32
assert value.shape[0] == nobj + 1
assert value.ndim == 2
pfeats[key] = value
logger.debug("merged, returning")
return all_features
def _extract(self, image, labels, atlas=None):
if not (image.ndim == labels.ndim == 4):
raise Exception("both images must be 4D. raw image shape: {}"
" label image shape: {}".format(image.shape, labels.shape))
# FIXME: maybe simplify? taggedShape should be easier here
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
#These are the feature names, selected by the user and the default feature names.
feature_names = deepcopy(self.Features([]).wait())
feature_names = self._augmentFeatureNames(feature_names)
# do global features
logger.debug("Computing global and default features")
global_features = {}
pool = RequestPool()
def compute_for_one_plugin(plugin_name, feature_dict):
plugin_inner = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
global_features[plugin_name] = plugin_inner.plugin_object.compute_global(image, labels, feature_dict, axes)
for plugin_name, feature_dict in feature_names.items():
if plugin_name != default_features_key:
pool.add(Request(partial(compute_for_one_plugin, plugin_name, feature_dict)))
pool.wait()
extrafeats = {}
for feat_key in default_features:
try:
sel = feature_names["Standard Object Features"][feat_key]["selected"]
except KeyError:
# we don't always set this property to True, sometimes it's just not there. The only important
# thing is that it's not False
sel = True
if not sel:
# This feature has not been selected by the user. Remove it from the computed dict into a special dict
# for default features
feature = global_features["Standard Object Features"].pop(feat_key)
else:
feature = global_features["Standard Object Features"][feat_key]
extrafeats[feat_key] = feature
if atlas is not None:
extrafeats['AtlasMapping'] = self._createAtlasMapping(extrafeats['RegionCenter'], atlas)
extrafeats = dict((k.replace(' ', ''), v)
for k, v in extrafeats.items())
mincoords = extrafeats["Coord<Minimum>"].astype(int)
maxcoords = extrafeats["Coord<Maximum>"].astype(int)
nobj = mincoords.shape[0]
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = collections.defaultdict(lambda: collections.defaultdict(list))
margin = max_margin(feature_names)
has_local_features = {}
for plugin_name, feature_dict in feature_names.items():
has_local_features[plugin_name] = False
for features in feature_dict.values():
if 'margin' in features:
has_local_features[plugin_name] = True
break
if numpy.any(margin) > 0:
#starting from 0, we stripped 0th background object in global computation
for i in range(0, nobj):
logger.debug("processing object {}".format(i))
extent = self.compute_extent(i, image, mincoords, maxcoords, axes, margin)
rawbbox = self.compute_rawbbox(image, extent, axes)
#it's i+1 here, because the background has label 0
binary_bbox = numpy.where(labels[tuple(extent)] == i+1, 1, 0).astype(numpy.bool)
for plugin_name, feature_dict in feature_names.items():
if not has_local_features[plugin_name]:
continue
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(rawbbox, binary_bbox, feature_dict, axes)
local_features[plugin_name] = dictextend(local_features[plugin_name], feats)
logger.debug("computing done, removing failures")
# remove local features that failed
for pname, pfeats in local_features.items():
for key in list(pfeats.keys()):
value = pfeats[key]
try:
pfeats[key] = numpy.vstack(list(v.reshape(1, -1) for v in value))
except:
logger.warning('feature {} failed'.format(key))
del pfeats[key]
# merge the global and local features
logger.debug("removed failed, merging")
all_features = {}
plugin_names = set(global_features.keys()) | set(local_features.keys())
for name in plugin_names:
d1 = global_features.get(name, {})
d2 = local_features.get(name, {})
all_features[name] = dict(list(d1.items()) + list(d2.items()))
all_features[default_features_key]=extrafeats
# reshape all features
for pfeats in all_features.values():
for key, value in pfeats.items():
if value.shape[0] != nobj:
raise Exception('feature {} does not have enough rows, {} instead of {}'.format(key, value.shape[0], nobj))
# because object classification operator expects nobj to
# include background. FIXME: we should change that assumption.
value = numpy.vstack((numpy.zeros(value.shape[1]),
value))
value = value.astype(numpy.float32) #turn Nones into numpy.NaNs
assert value.dtype == numpy.float32
assert value.shape[0] == nobj+1
assert value.ndim == 2
pfeats[key] = value
logger.debug("merged, returning")
return all_features
def compute_for_one_plugin(plugin_name, feature_dict):
plugin_inner = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
global_features[plugin_name] = plugin_inner.plugin_object.compute_global(image, labels, feature_dict, axes)
def flatten_ilastik_feature_table(table, selection, signal):
selection = list(selection)
frames = table.meta.shape[0]
logger.info("Fetching object features for feature table...")
computed_feature = table([]).wait()
signal(0)
feature_long_names = [] # For example, "Size in Pixels"
feature_short_names = [] # For example, "Count"
feature_plugins = []
feature_channels = []
feature_types = []
for plugin_name, feature_dict in computed_feature[0].items():
all_props = None
if plugin_name == default_features_key:
plugin = pluginManager.getPluginByName("Standard Object Features",
"ObjectFeatures")
else:
plugin = pluginManager.getPluginByName(plugin_name,
"ObjectFeatures")
if plugin:
plugin_feature_names = {el: {} for el in list(feature_dict.keys())}
all_props = plugin.plugin_object.fill_properties(
plugin_feature_names) # fill in display name and such
for feat_name, feat_array in feature_dict.items():
if all_props:
long_name = all_props[feat_name]["displaytext"]
else:
long_name = feat_name
if (plugin_name == default_features_key or long_name in selection
or feat_name
in selection) and long_name not in feature_long_names:
feature_long_names.append(long_name)
feature_short_names.append(feat_name)
feature_plugins.append(plugin_name)
feature_channels.append((feat_array.shape[1]))
feature_types.append(feat_array.dtype)
signal(25)
obj_count = []
for t, cf in computed_feature.items():
obj_count.append(cf[default_features_key]["Count"].shape[0] -
1) # no background
signal(50)
dtype_names = []
dtype_types = []
dtype_to_key = {}
for i, name in enumerate(feature_long_names):
if feature_channels[i] > 1:
for c in range(feature_channels[i]):
dtype_names.append("%s_%i" % (name, c))
dtype_types.append(feature_types[i].name)
dtype_to_key[dtype_names[-1]] = (feature_plugins[i],
feature_short_names[i], c)
else:
dtype_names.append(name)
dtype_types.append(feature_types[i].name)
dtype_to_key[dtype_names[-1]] = (feature_plugins[i],
feature_short_names[i], 0)
feature_table = np.zeros((sum(obj_count), ), dtype=",".join(dtype_types))
feature_table.dtype.names = list(map(str, dtype_names))
signal(75)
start = 0
end = obj_count[0]
for t, cf in computed_feature.items():
for name in dtype_names:
plugin, feat_name, index = dtype_to_key[name]
data_len = len(cf[plugin][feat_name][1:, index])
feature_table[name][start:start +
data_len] = cf[plugin][feat_name][1:, index]
start = end
try:
end += obj_count[int(t) + 1]
except IndexError:
end = sum(obj_count)
signal(100)
return feature_table
def _extract(self, image, labels):
if not (image.ndim == labels.ndim == 4):
raise Exception("both images must be 4D. raw image shape: {}"
" label image shape: {}".format(image.shape, labels.shape))
# FIXME: maybe simplify? taggedShape should be easier here
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
logger.debug("Computing default features")
feature_names = deepcopy(self.Features([]).wait())
# do global features
logger.debug("computing global features")
extra_features_computed = False
global_features = {}
selected_vigra_features = []
for plugin_name, feature_dict in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
if plugin_name == "Standard Object Features":
#expand the feature list by our default features
logger.debug("attaching default features {} to vigra features {}".format(default_features, feature_dict))
selected_vigra_features = feature_dict.keys()
feature_dict.update(default_features)
extra_features_computed = True
global_features[plugin_name] = plugin.plugin_object.compute_global(image, labels, feature_dict, axes)
extrafeats = {}
if extra_features_computed:
for feat_key in default_features:
feature = None
if feat_key in selected_vigra_features:
#we wanted that feature independently
feature = global_features["Standard Object Features"][feat_key]
else:
feature = global_features["Standard Object Features"].pop(feat_key)
feature_names["Standard Object Features"].pop(feat_key)
extrafeats[feat_key] = feature
else:
logger.debug("default features not computed, computing separately")
extrafeats_acc = vigra.analysis.extractRegionFeatures(image[slc3d].squeeze().astype(np.float32), labels.squeeze(),
default_features.keys(),
ignoreLabel=0)
#remove the 0th object, we'll add it again later
for k, v in extrafeats_acc.iteritems():
extrafeats[k]=v[1:]
if len(v.shape)==1:
extrafeats[k]=extrafeats[k].reshape(extrafeats[k].shape+(1,))
extrafeats = dict((k.replace(' ', ''), v)
for k, v in extrafeats.iteritems())
mincoords = extrafeats["Coord<Minimum>"]
maxcoords = extrafeats["Coord<Maximum>"]
nobj = mincoords.shape[0]
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = collections.defaultdict(lambda: collections.defaultdict(list))
margin = max_margin(feature_names)
has_local_features = {}
for plugin_name, feature_dict in feature_names.iteritems():
has_local_features[plugin_name] = False
for features in feature_dict.itervalues():
if 'margin' in features:
has_local_features[plugin_name] = True
break
if np.any(margin) > 0:
#starting from 0, we stripped 0th background object in global computation
for i in range(0, nobj):
logger.debug("processing object {}".format(i))
extent = self.compute_extent(i, image, mincoords, maxcoords, axes, margin)
rawbbox = self.compute_rawbbox(image, extent, axes)
#it's i+1 here, because the background has label 0
binary_bbox = np.where(labels[tuple(extent)] == i+1, 1, 0).astype(np.bool)
for plugin_name, feature_dict in feature_names.iteritems():
if not has_local_features[plugin_name]:
continue
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(rawbbox, binary_bbox, feature_dict, axes)
local_features[plugin_name] = dictextend(local_features[plugin_name], feats)
logger.debug("computing done, removing failures")
# remove local features that failed
for pname, pfeats in local_features.iteritems():
for key in pfeats.keys():
value = pfeats[key]
try:
pfeats[key] = np.vstack(list(v.reshape(1, -1) for v in value))
except:
logger.warn('feature {} failed'.format(key))
del pfeats[key]
# merge the global and local features
logger.debug("removed failed, merging")
all_features = {}
plugin_names = set(global_features.keys()) | set(local_features.keys())
for name in plugin_names:
d1 = global_features.get(name, {})
d2 = local_features.get(name, {})
all_features[name] = dict(d1.items() + d2.items())
all_features[default_features_key]=extrafeats
# reshape all features
for pfeats in all_features.itervalues():
for key, value in pfeats.iteritems():
if value.shape[0] != nobj:
raise Exception('feature {} does not have enough rows, {} instead of {}'.format(key, value.shape[0], nobj))
# because object classification operator expects nobj to
# include background. FIXME: we should change that assumption.
value = np.vstack((np.zeros(value.shape[1]),
value))
value = value.astype(np.float32) #turn Nones into numpy.NaNs
assert value.dtype == np.float32
assert value.shape[0] == nobj+1
assert value.ndim == 2
pfeats[key] = value
logger.debug("merged, returning")
return all_features
def _initMetaInfoText(self):
## meta-info display widgets
plugin = pluginManager.getPluginByName(self.pluginName, category="TrackingExportFormats")
self.metaInfoTextEdit.setHtml(plugin.description)
def _extract(self, image, labels):
assert image.ndim == labels.ndim == 4, "Images must be 4D. Shapes were: {} and {}".format(image.shape, labels.shape)
class Axes(object):
x = image.axistags.index('x')
y = image.axistags.index('y')
z = image.axistags.index('z')
c = image.axistags.index('c')
axes = Axes()
image = np.asarray(image, dtype=np.float32)
labels = np.asarray(labels, dtype=np.uint32)
slc3d = [slice(None)] * 4 # FIXME: do not hardcode
slc3d[axes.c] = 0
labels = labels[slc3d]
extrafeats = vigra.analysis.extractRegionFeatures(image[slc3d], labels,
gui_features,
ignoreLabel=0)
mincoords = extrafeats["Coord<Minimum >"]
maxcoords = extrafeats["Coord<Maximum >"]
nobj = mincoords.shape[0]
feature_names = self.Features([]).wait()
# do global features
global_features = defaultdict(list)
for plugin_name, feature_list in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_global(image, labels, feature_list, axes)
global_features = dict(global_features.items() + feats.items())
# local features: loop over all objects
def dictextend(a, b):
for key in b:
a[key].append(b[key])
return a
local_features = defaultdict(list)
for i in range(1, nobj):
print "processing object {}".format(i)
mins, maxs = self.compute_minmax(i, image, mincoords, maxcoords, axes)
rawbbox = self.compute_rawbbox(image, mins, maxs, axes)
label_bboxes = self.compute_label_bboxes(i, labels, mins, maxs, axes)
for plugin_name, feature_list in feature_names.iteritems():
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
feats = plugin.plugin_object.compute_local(rawbbox, label_bboxes, feature_list, axes, mins, maxs)
local_features = dictextend(local_features, feats)
for key in local_features.keys():
value = local_features[key]
try:
local_features[key] = np.vstack(list(v.reshape(1, -1) for v in value))
except:
print 'warning: feature {} failed'.format(key)
del local_features[key]
all_features = dict(global_features.items() + local_features.items())
for key, value in all_features.iteritems():
if value.shape[0] != nobj - 1:
raise Exception('feature {} does not have enough rows')
# because object classification operator expects nobj to
# include background. we should change that assumption.
value = np.vstack((np.zeros(value.shape[1]),
value))
value = value.astype(np.float32)
assert value.dtype == np.float32
assert value.shape[0] == nobj
assert value.ndim == 2
all_features[key] = value
# add features needed by downstream applets. these should be
# removed before classification.
extrafeats = dict((k.replace(' ', '') + gui_features_suffix, v)
for k, v in extrafeats.iteritems())
return dict(all_features.items() + extrafeats.items())
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# FIXME: This probably only works for the non-blockwise export slot.
# We should assert that the user isn't using the blockwise slot.
# Plugin export if selected
logger.info("Export source is: " + self.dataExportTrackingApplet.
topLevelOperator.SelectedExportSource.value)
print "in post_process_lane_export"
if self.dataExportTrackingApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportTrackingApplet.topLevelOperator.SelectedPlugin.value
exportPluginInfo = pluginManager.getPluginByName(
selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" %
exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" %
selectedPlugin)
name_format = self.dataExportTrackingApplet.topLevelOperator.getLane(
lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(
lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = os.path.dirname(partially_formatted_name)
if filename is None or len(str(filename)) == 0:
logger.error(
"Cannot export from plugin with empty output filename")
return
exportStatus = self.trackingApplet.topLevelOperator.getLane(
lane_index).exportPlugin(filename, exportPlugin,
checkOverwriteFiles)
if not exportStatus:
return False
logger.info("Export done")
return
# CSV Table export (only if plugin was not selected)
settings, selected_features = self.trackingApplet.topLevelOperator.getLane(
lane_index).get_table_export_settings()
from lazyflow.utility import PathComponents, make_absolute, format_known_keys
if settings:
self.dataExportTrackingApplet.progressSignal.emit(-1)
raw_dataset_info = self.dataSelectionApplet.topLevelOperator.DatasetGroup[
lane_index][0].value
project_path = self.shell.projectManager.currentProjectPath
project_dir = os.path.dirname(project_path)
dataset_dir = PathComponents(
raw_dataset_info.filePath).externalDirectory
abs_dataset_dir = make_absolute(dataset_dir, cwd=project_dir)
known_keys = {}
known_keys['dataset_dir'] = abs_dataset_dir
nickname = raw_dataset_info.nickname.replace('*', '')
if os.path.pathsep in nickname:
nickname = PathComponents(nickname.split(
os.path.pathsep)[0]).fileNameBase
known_keys['nickname'] = nickname
# use partial formatting to fill in non-coordinate name fields
name_format = settings['file path']
partially_formatted_name = format_known_keys(
name_format, known_keys)
settings['file path'] = partially_formatted_name
req = self.trackingApplet.topLevelOperator.getLane(
lane_index
).export_object_data(
lane_index,
# FIXME: Even in non-headless mode, we can't show the gui because we're running in a non-main thread.
# That's not a huge deal, because there's still a progress bar for the overall export.
show_gui=False)
req.wait()
self.dataExportTrackingApplet.progressSignal.emit(100)
def post_process_lane_export(self, lane_index, checkOverwriteFiles=False):
# `time` parameter ensures we check only once for files that could be overwritten, pop up
# the MessageBox and then don't export (time=0). For the next round we click the export button,
# we really want it to export, so time=1. The default parameter is 1, so everything but not 0,
# in order to ensure writing out even in headless mode.
# FIXME: This probably only works for the non-blockwise export slot.
# We should assert that the user isn't using the blockwise slot.
# Plugin export if selected
logger.info(
"Export source is: " +
self.dataExportApplet.topLevelOperator.SelectedExportSource.value)
if self.dataExportApplet.topLevelOperator.SelectedExportSource.value == OpTrackingBaseDataExport.PluginOnlyName:
logger.info("Export source plugin selected!")
selectedPlugin = self.dataExportApplet.topLevelOperator.SelectedPlugin.value
exportPluginInfo = pluginManager.getPluginByName(
selectedPlugin, category="TrackingExportFormats")
if exportPluginInfo is None:
logger.error("Could not find selected plugin %s" %
exportPluginInfo)
else:
exportPlugin = exportPluginInfo.plugin_object
logger.info("Exporting tracking result using %s" %
selectedPlugin)
name_format = self.dataExportApplet.topLevelOperator.getLane(
lane_index).OutputFilenameFormat.value
partially_formatted_name = self.getPartiallyFormattedName(
lane_index, name_format)
if exportPlugin.exportsToFile:
filename = partially_formatted_name
if os.path.basename(filename) == '':
filename = os.path.join(filename, 'pluginExport.txt')
else:
filename = os.path.dirname(partially_formatted_name)
if filename is None or len(str(filename)) == 0:
logger.error(
"Cannot export from plugin with empty output filename")
return
exportStatus = self.trackingApplet.topLevelOperator.getLane(
lane_index).exportPlugin(filename, exportPlugin,
checkOverwriteFiles)
if not exportStatus:
return False
logger.info("Export done")
return
# CSV Table export (only if plugin was not selected)
settings, selected_features = self.trackingApplet.topLevelOperator.getLane(
lane_index).get_table_export_settings()
if settings:
self.dataExportApplet.progressSignal.emit(0)
name_format = settings['file path']
partially_formatted_name = self.getPartiallyFormattedName(
lane_index, name_format)
settings['file path'] = partially_formatted_name
req = self.trackingApplet.topLevelOperator.getLane(
lane_index
).export_object_data(
lane_index,
# FIXME: Even in non-headless mode, we can't show the gui because we're running in a non-main thread.
# That's not a huge deal, because there's still a progress bar for the overall export.
show_gui=False)
req.wait()
self.dataExportApplet.progressSignal.emit(100)
# Restore option to bypass cache to false
self.objectExtractionApplet.topLevelOperator[
lane_index].BypassModeEnabled.setValue(False)
# Restore state of axis ranges
parameters = self.trackingApplet.topLevelOperator.Parameters.value
parameters['time_range'] = self.prev_time_range
parameters['x_range'] = self.prev_x_range
parameters['y_range'] = self.prev_y_range
parameters['z_range'] = self.prev_z_range
def _initMetaInfoText(self):
## meta-info display widgets
plugin = pluginManager.getPluginByName(
self.pluginName, category="TrackingExportFormats")
self.metaInfoTextEdit.setHtml(plugin.description)
def populate(self):
#self.ui.treeWidget.setColumnCount(2)
for pluginName, features in self.featureDict.items():
if pluginName == "TestFeatures" and not ilastik_config.getboolean(
"ilastik", "debug"):
continue
parent = QTreeWidgetItem(self.ui.treeWidget)
parent.setText(0, pluginName)
parent.setFlags(Qt.ItemIsUserCheckable | Qt.ItemIsEnabled)
# hack to ensure checkboxes visible
parent.setCheckState(0, Qt.Checked)
parent.setCheckState(0, Qt.Unchecked)
parent.setExpanded(False)
self.countChecked[pluginName] = 0
self.countAll[pluginName] = len(self.featureDict[pluginName])
advanced_names = []
simple_names = []
selected_names = []
groups = set()
plugin = pluginManager.getPluginByName(pluginName,
"ObjectFeatures")
features_with_props = deepcopy(features)
if plugin is not None:
plugin.plugin_object.fill_properties(features_with_props)
for name in sorted(features.keys()):
parameters = features[name]
for prop, prop_value in features_with_props[name].items():
if not prop in list(parameters.keys()):
# this property has not been added yet (perhaps the feature dictionary has been read from file)
# set it now
parameters[prop] = prop_value
try:
if parameters['advanced'] is True:
advanced_names.append(name)
else:
simple_names.append(name)
except KeyError:
simple_names.append(name)
try:
groups.add(parameters["group"])
except KeyError:
pass
if pluginName in self.selectedFeatures:
if name in self.selectedFeatures[pluginName]:
selected_names.append(name)
gr_items = {}
for gr in groups:
gr_items[gr] = QTreeWidgetItem(parent)
gr_items[gr].setText(0, gr)
#gr_items[gr].setFlags(Qt.ItemIsEnabled | Qt.ItemIsUserCheckable)
gr_items[gr].setExpanded(True)
for name in simple_names + advanced_names:
if name in advanced_names and (not name in selected_names):
# do not display advanced features, if they have not been selected previously
continue
parameters = features[name]
if "group" in parameters:
item = QTreeWidgetItem(gr_items[parameters["group"]])
item.group_name = parameters["group"]
else:
item = QTreeWidgetItem(parent)
if 'displaytext' in parameters:
itemtext = parameters['displaytext']
else:
itemtext = name
item.setText(0, itemtext)
item.feature_id = name
item.setFlags(Qt.ItemIsUserCheckable | Qt.ItemIsEnabled)
if 'tooltip' in parameters:
item.setToolTip(0, parameters['tooltip'])
# hack to ensure checkboxes visible
item.setCheckState(0, Qt.Checked)
item.setCheckState(0, Qt.Unchecked)
if name in selected_names:
item.setCheckState(0, Qt.Checked)
self.countChecked[pluginName] += 1
if self.countChecked[pluginName] == 0:
parent.setCheckState(0, Qt.Unchecked)
elif self.countChecked[pluginName] == self.countAll[pluginName]:
parent.setCheckState(0, Qt.Checked)
else:
parent.setCheckState(0, Qt.PartiallyChecked)
self.updateToolTip(parent)
# facilitates switching of the CheckBox when clicking on the Text of a QTreeWidgetItem
self.ui.treeWidget.setCurrentItem(None)
def flatten_ilastik_feature_table(table, selection, signal):
selection = list(selection)
frames = table.meta.shape[0]
logger.info('Fetching object features for feature table...')
computed_feature = table([]).wait()
signal(0)
feature_long_names = [] # For example, "Size in Pixels"
feature_short_names = [] # For example, "Count"
feature_plugins = []
feature_channels = []
feature_types = []
for plugin_name, feature_dict in computed_feature[0].items():
all_props = None
if plugin_name==default_features_key:
plugin = pluginManager.getPluginByName("Standard Object Features", "ObjectFeatures")
else:
plugin = pluginManager.getPluginByName(plugin_name, "ObjectFeatures")
if plugin:
plugin_feature_names = {el:{} for el in list(feature_dict.keys())}
all_props = plugin.plugin_object.fill_properties(plugin_feature_names) #fill in display name and such
for feat_name, feat_array in feature_dict.items():
if all_props:
long_name = all_props[feat_name]["displaytext"]
else:
long_name = feat_name
if (plugin_name == default_features_key or \
long_name in selection or \
feat_name in selection) and \
long_name not in feature_long_names:
feature_long_names.append(long_name)
feature_short_names.append(feat_name)
feature_plugins.append(plugin_name)
feature_channels.append((feat_array.shape[1]))
feature_types.append(feat_array.dtype)
signal(25)
obj_count = []
for t, cf in computed_feature.items():
obj_count.append(cf[default_features_key]["Count"].shape[0] - 1) # no background
signal(50)
dtype_names = []
dtype_types = []
dtype_to_key = {}
for i, name in enumerate(feature_long_names):
if feature_channels[i] > 1:
for c in range(feature_channels[i]):
dtype_names.append("%s_%i" % (name, c))
dtype_types.append(feature_types[i].name)
dtype_to_key[dtype_names[-1]] = (feature_plugins[i], feature_short_names[i], c)
else:
dtype_names.append(name)
dtype_types.append(feature_types[i].name)
dtype_to_key[dtype_names[-1]] = (feature_plugins[i], feature_short_names[i], 0)
feature_table = np.zeros((sum(obj_count),), dtype=",".join(dtype_types))
feature_table.dtype.names = list(map(str, dtype_names))
signal(75)
start = 0
end = obj_count[0]
for t, cf in computed_feature.items():
for name in dtype_names:
plugin, feat_name, index = dtype_to_key[name]
data_len = len(cf[plugin][feat_name][1:, index])
feature_table[name][start:start + data_len] = cf[plugin][feat_name][1:, index]
start = end
try:
end += obj_count[int(t) + 1]
except IndexError:
end = sum(obj_count)
signal(100)
return feature_table