added ProbabilityMapper to GUI

napari_accelerated_pixel_and_object_classification/_dock_widget.py

@@ -17,7 +17,7 @@
 import numpy as np
 import napari
 
-from apoc import PredefinedFeatureSet, ObjectSegmenter, PixelClassifier
+from apoc import PredefinedFeatureSet, ObjectSegmenter, PixelClassifier, ProbabilityMapper
 from napari_tools_menu import register_dock_widget
 
 @register_dock_widget(menu="Segmentation > Object segmentation (APOC)")
@@ -61,7 +61,13 @@ def __init__(self, napari_viewer, classifier_class=ObjectSegmenter):
         training_widget.setLayout(QVBoxLayout())
 
         # Annotation
-        training_widget.layout().addWidget(QLabel("Select ground truth annotation / object label"))
+        if self.classifier_class == ObjectSegmentation:
+            suffix = " + object class"
+        elif self.classifier_class == ProbabilityMapper:
+            suffix = " + class for probability output"
+        else:
+            suffix = ""
+        training_widget.layout().addWidget(QLabel("Select ground truth annotation" + suffix))
         self.label_list = QComboBox()
         self.update_label_list()
 
@@ -74,9 +80,11 @@ def __init__(self, napari_viewer, classifier_class=ObjectSegmenter):
         num_object_annotation_spinner.setMaximumWidth(40)
         num_object_annotation_spinner.setMinimum(1)
         num_object_annotation_spinner.setValue(2)
-        if self.classifier_class != PixelClassifier:
-            print("classifier ", self.classifier_class)
+        if self.classifier_class == ObjectSegmentation:
             temp.layout().addWidget(num_object_annotation_spinner)
+        elif self.classifier_class == ProbabilityMapper:
+            temp.layout().addWidget(num_object_annotation_spinner)
+
         training_widget.layout().addWidget(temp)
 
         # Features
@@ -217,8 +225,10 @@ def train(self, images, annotation, object_annotation_value, feature_definition,
             num_ensembles=num_trees,
             max_depth=num_max_depth)
 
-        if hasattr(clf, "positive_class_identifier"):
+        if self.classifier_class == ObjectSegmenter:
             clf.positive_class_identifier = object_annotation_value
+        elif self.classifier_class == ProbabilityMapper:
+            clf.output_probability_of_class = object_annotation_value
 
         print("annotation shape", annotation.shape)
 
@@ -238,7 +248,10 @@ def _add_to_viewer(self, name, data):
             self.viewer.layers[name].data = data
             self.viewer.layers[name].visible = True
         except KeyError:
-            self.viewer.add_labels(data, name=name)
+            if self.classifier_class == ProbabilityMapper:
+                self.viewer.add_image(data, name=name)
+            else:
+                self.viewer.add_labels(data, name=name)
 
     def predict(self, images, filename):
         print("predict")
@@ -375,6 +388,12 @@ class SemanticSegmentation(ObjectSegmentation):
     def __init__(self, napari_viewer):
         super().__init__(napari_viewer, classifier_class=PixelClassifier)
 
+@register_dock_widget(menu="Filtering > Probability mapper (APOC)")
+class ProbabilityMapping(ObjectSegmentation):
+    def __init__(self, napari_viewer):
+        super().__init__(napari_viewer, classifier_class=ProbabilityMapper)
+
+
 class FeatureSelector(QWidget):
     def __init__(self, feature_definition:str):
         super().__init__()

napari_accelerated_pixel_and_object_classification/_function.py

@@ -1,4 +1,4 @@
-from apoc import PredefinedFeatureSet, PixelClassifier, ObjectSegmenter, ObjectClassifier
+from apoc import PredefinedFeatureSet, PixelClassifier, ObjectSegmenter, ObjectClassifier, ProbabilityMapper
 
 import numpy as np
 from napari_plugin_engine import napari_hook_implementation
@@ -12,6 +12,7 @@ def napari_experimental_provide_function():
     return [
         Train_object_segmentation,
         Apply_object_segmentation,
+        Apply_probability_mapper,
         Train_object_segmentation_from_visible_image_layers,
         Apply_object_segmentation_to_visible_image_layers,
         Train_pixel_classifier,
@@ -101,6 +102,16 @@ def Train_object_segmentation(
     return result
 
 
+@register_function(menu="Filtering > Probability Mapper (apply pretrained, APOC)")
+@time_slicer
+def Apply_probability_mapper(image: "napari.types.ImageData",
+                              model_filename : str = "ProbabilityMapper.cl",
+                              viewer: napari.Viewer = None) -> "napari.types.ImageData":
+    clf = ProbabilityMapper(opencl_filename=model_filename)
+    result = clf.predict(image=image)
+    return result
+
+
 @register_function(menu="Segmentation > Object segmentation (apply pretrained, APOC)")
 @time_slicer
 def Apply_object_segmentation(image: "napari.types.ImageData",