Package qupath.opencv.ml.objects.features

Interface FeatureExtractor<T>

Type Parameters:

T - generic parameter related to the ImageData (not always required).

public interface FeatureExtractor<T>

Interface for extracting features from PathObjects for the purpose of object classification.

Author:

Pete Bankhead

See Also:

• ObjectClassifier

Method Summary

Modifier and Type	Method	Description
void	extractFeatures(ImageData<T> imageData, Collection<? extends PathObject> pathObjects, FloatBuffer buffer)	Extract features from a collection of objects, adding the corresponding values to a FloatBuffer.
List<String>	getFeatureNames()
Collection<String>	getMissingFeatures(ImageData<T> imageData, PathObject pathObject)	Check for missing features, returning the names.
int	nFeatures()

Method Details

getFeatureNames

List<String> getFeatureNames()

Returns:

the names of the features that can be extracted

nFeatures

int nFeatures()

Returns:

the number of features that can be extracted

extractFeatures

void extractFeatures(ImageData<T> imageData,
 Collection<? extends PathObject> pathObjects,
 FloatBuffer buffer)

Extract features from a collection of objects, adding the corresponding values to a FloatBuffer. Each feature is a single numeric value. At the end, pathObjects.size() * nFeatures() features should be added to the buffer. Each object is processed in turn, with all features for that object added before the next object is processed.

Parameters:

imageData - image data, used by some implementations to determine feature values (e.g. if these depend upon pixels).

pathObjects - objects for which features should be calculated

buffer - buffer into which feature values should be added

getMissingFeatures

Collection<String> getMissingFeatures(ImageData<T> imageData,
 PathObject pathObject)

Check for missing features, returning the names. This is useful as a warning that the input for the feature extractor may not be valid. Default implementation returns an empty list; however, implementations should attempt to provide a meaningful output if possible.

Parameters:

imageData - image containing the objects to test

pathObject - object to test for missing feature

Returns:

a collection of feature names that correspond to missing features