Package boofcv.alg.tracker.tld

Class TldDetection<T extends ImageGray<T>>

java.lang.Object

boofcv.alg.tracker.tld.TldDetection<T>

public class TldDetection<T extends ImageGray<T>>
extends Object

Runs a detection cascade for each region. The cascade is composed of a variance test, Fern classifier, and finally the template classifier. The next test in the cascade is only considered if the previous passes. Several changes have been made in how the cascade operates compared to the original paper. See code for comments.

Field Summary

Fields

Modifier and Type	Field	Description
protected boolean	ambiguous
protected DogArray<TldRegion>	candidateDetections
protected ConfigTld	config
protected DogArray<TldRegionFernInfo>	fernInfo
protected List<ImageRectangle>	fernRegions
protected TldTemplateMatching<T>	template

Constructor Summary

Constructors

Modifier	Constructor	Description
protected	TldDetection()
	TldDetection(TldFernClassifier<T> fern, TldTemplateMatching<T> template, TldVarianceFilter<T> variance, ConfigTld config)

Method Summary

Modifier and Type	Method	Description
protected void	computeTemplateConfidence()	Computes the confidence for all the regions which pass the fern test
protected void	detectionCascade(DogArray<ImageRectangle> cascadeRegions)	Detects the object inside the image.
List<ImageRectangle>	getAmbiguousRegions()
@Nullable TldRegion	getBest()
DogArray<TldRegion>	getCandidateDetections()
DogArray<TldRegionFernInfo>	getFernInfo()
DogArray<TldRegion>	getLocalMaximums()
TldNonMaximalSuppression	getNonmax()
List<ImageRectangle>	getSelectedFernRectangles()	Rectangles selected by the fern classifier as candidates
DogArray_F64	getStorageMetric()
List<ImageRectangle>	getStorageRect()
boolean	isAmbiguous()
boolean	isSuccess()
@Nullable TldRegion	selectBest()
protected void	selectBestRegionsFern(double totalP, double totalN)	compute the probability that each region is the target conditional upon this image the sumP and sumN are needed for image conditional probability NOTE: This is a big change from the original paper

Methods inherited from class java.lang.Object

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

Field Details

template

protected TldTemplateMatching<T extends ImageGray<T>> template

fernInfo

protected DogArray<TldRegionFernInfo> fernInfo

config

protected ConfigTld config

fernRegions

protected List<ImageRectangle> fernRegions

candidateDetections

protected DogArray<TldRegion> candidateDetections

ambiguous

protected boolean ambiguous

Constructor Details

TldDetection

public TldDetection(TldFernClassifier<T> fern,
 TldTemplateMatching<T> template,
 TldVarianceFilter<T> variance,
 ConfigTld config)

TldDetection

protected TldDetection()

Method Details

detectionCascade

protected void detectionCascade(DogArray<ImageRectangle> cascadeRegions)

Detects the object inside the image. Eliminates candidate regions using a cascade of tests

computeTemplateConfidence

protected void computeTemplateConfidence()

Computes the confidence for all the regions which pass the fern test

selectBestRegionsFern

protected void selectBestRegionsFern(double totalP,
 double totalN)

compute the probability that each region is the target conditional upon this image the sumP and sumN are needed for image conditional probability NOTE: This is a big change from the original paper

selectBest

@Nullable
public @Nullable TldRegion selectBest()

getBest

@Nullable
public @Nullable TldRegion getBest()

isAmbiguous

public boolean isAmbiguous()

getNonmax

public TldNonMaximalSuppression getNonmax()

getStorageMetric

public DogArray_F64 getStorageMetric()

getStorageRect

public List<ImageRectangle> getStorageRect()

getCandidateDetections

public DogArray<TldRegion> getCandidateDetections()

getLocalMaximums

public DogArray<TldRegion> getLocalMaximums()

getAmbiguousRegions

public List<ImageRectangle> getAmbiguousRegions()

getFernInfo

public DogArray<TldRegionFernInfo> getFernInfo()

getSelectedFernRectangles

public List<ImageRectangle> getSelectedFernRectangles()

Rectangles selected by the fern classifier as candidates

Returns:

List of rectangles

isSuccess

public boolean isSuccess()