Package qupath.opencv.ml.objects

Class OpenCVMLClassifier<T>

java.lang.Object

qupath.lib.classifiers.object.AbstractObjectClassifier<T>

qupath.opencv.ml.objects.OpenCVMLClassifier<T>

Type Parameters:

T -

All Implemented Interfaces:

ObjectClassifier<T>

public class OpenCVMLClassifier<T>
extends AbstractObjectClassifier<T>

An ObjectClassifier that uses an OpenCVClassifiers.OpenCVStatModel for classification.

Author:

Pete Bankhead

Field Summary

Fields inherited from interface qupath.lib.classifiers.object.ObjectClassifier

PROJECT_LOCATION

Method Summary

Modifier and Type	Method	Description
int	classifyObjects(ImageData<T> imageData, Collection<? extends PathObject> pathObjects, boolean resetExistingClass)	Classify the objects in the specified collection.
static <T> ObjectClassifier<T>	create(OpenCVClassifiers.OpenCVStatModel model, PathObjectFilter filter, FeatureExtractor<T> extractor, List<PathClass> pathClasses)	Create a new ObjectClassifier that uses an OpenCVClassifiers.OpenCVStatModel for classification.
Map<String,Integer>	getMissingFeatures(ImageData<T> imageData, Collection<? extends PathObject> pathObjects)	Check for missing features, returning the names and number of input objects missing the specified features.
Collection<PathClass>	getPathClasses()	Get the classifications set by this classifier.
String	toString()

Methods inherited from class qupath.lib.classifiers.object.AbstractObjectClassifier

classifyObjects, getCompatibleObjects

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, wait, wait, wait

Method Details

create

public static <T> ObjectClassifier<T> create(OpenCVClassifiers.OpenCVStatModel model,
 PathObjectFilter filter,
 FeatureExtractor<T> extractor,
 List<PathClass> pathClasses)

Create a new ObjectClassifier that uses an OpenCVClassifiers.OpenCVStatModel for classification.

Type Parameters:

T - generic type, which matches that of an ImageData

Parameters:

model - the OpenCVClassifiers.OpenCVStatModel used to apply the prediction

filter - a filter used to select objects from a hierarchy

extractor - a feature extractor to determine features for each object

pathClasses - available classifications; the order is important, and relates to the classification output

Returns:

getPathClasses

public Collection<PathClass> getPathClasses()

Description copied from interface: ObjectClassifier

Get the classifications set by this classifier.

Returns:

classifyObjects

public int classifyObjects(ImageData<T> imageData,
 Collection<? extends PathObject> pathObjects,
 boolean resetExistingClass)

Description copied from interface: ObjectClassifier

Classify the objects in the specified collection. This provides a means to specify exactly which objects should be classified, avoiding reliance on ObjectClassifier.getCompatibleObjects(ImageData).

Note: this method does not typically fire any hierarchy change/object classification events. It is up to the caller to fire these events if required.

Parameters:

imageData - data that may or may not be required for classification depending upon how features are extracted

pathObjects - the objects to classify

resetExistingClass -

Returns:

the number of objects whose classification was changed.

See Also:

• ObjectClassifier.classifyObjects(ImageData, boolean)
• ObjectClassifier.getCompatibleObjects(ImageData)

toString

public String toString()

Overrides:

toString in class Object

getMissingFeatures

public Map<String,Integer> getMissingFeatures(ImageData<T> imageData,
 Collection<? extends PathObject> pathObjects)

Description copied from interface: ObjectClassifier

Check for missing features, returning the names and number of input objects missing the specified features. This is useful as a warning that the input for the classifier may not be valid. Default implementation returns an empty map; however, implementations should attempt to provide a meaningful output if possible. Features that are not missing should not be included in the output.

Parameters:

imageData - image containing the objects to test

pathObjects - objects to test for missing features; if not available, ObjectClassifier.getCompatibleObjects(ImageData) will be called.

Returns:

a map of feature names and the number of objects missing the corresponding features.