Collection of metrics used to evaluate the results of a ELocator tool. More...

#include <eVision/EasyDeepLearning/EasyLocate/ELocatorMetrics.h>

Collaboration diagram for eVision::EasyDeepLearning::ELocatorMetrics:

Public Member Functions
	ELocatorMetrics ()
	Constructs an ELocatorMetrics object.
	ELocatorMetrics (const eVision::EasyDeepLearning::ELocatorMetrics &other)
	Copy constructor of an ELocatorMetrics object.
virtual	~ELocatorMetrics () throw ()
eVision::EasyDeepLearning::ELocatorMetrics &	operator= (const eVision::EasyDeepLearning::ELocatorMetrics &other)
	Assignment operator.
void	Load (const std::string &path)
	Loads a locator metric. The given ESerializer must have been created for reading.
void	Load (eVision::ESerializer *serializer)
	Loads a locator metric. The given ESerializer must have been created for reading.
void	Save (const std::string &path)
	Saves a locator metric. The given ESerializer must have been created for writing.
void	Save (eVision::ESerializer *serializer)
	Saves a locator metric. The given ESerializer must have been created for writing.
bool	IsValid () const
	Indicates whether the object contains at least one result.
float	GetError () const
	Error of the EasyLocate training algorithm (only available for the training metrics, see ELocatorBase::GetTrainingMetrics).
float	GetRecall () const
	Recall (true positive rate, proportion of ground truth object that are correctly detected).
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetWeightedPrecisionRecallPoint () const
	Weighted precision and recall metrics for the current or specified threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetWeightedPrecisionRecallPoint (float threshold) const
	Weighted precision and recall metrics for the current or specified threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetWeightedPrecisionRecallPoint (const std::vector< float > &weights) const
	Weighted precision and recall metrics for the current or specified threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetWeightedPrecisionRecallPoint (float threshold, const std::vector< float > &weights) const
	Weighted precision and recall metrics for the current or specified threshold.
float	GetWeightedRecall () const
	Weighted average of the recall (true positive rate) for each label.
float	GetWeightedRecall (const std::vector< float > &weights) const
	Weighted average of the recall (true positive rate) for each label.
float	GetLabelRecall (const std::string &label) const
	Recall for the given label.
float	GetLabelRecall (int labelId) const
	Recall for the given label.
float	GetBestWeightedRecall () const
	Best weighted recall achievable by changing the detection threshold.
float	GetBestWeightedRecall (const std::vector< float > &weights) const
	Best weighted recall achievable by changing the detection threshold.
float	GetBestWeightedRecallThreshold () const
	Detection threshold that gives the best weighted recall (ELocatorMetrics::GetWeightedRecall).
float	GetBestWeightedRecallThreshold (const std::vector< float > &weights) const
	Detection threshold that gives the best weighted recall (ELocatorMetrics::GetWeightedRecall).
void	GetBestWeightedRecallAndThreshold (float &recall, float &threshold) const
	Best weighted recall achievable with the corresponding threshold.
void	GetBestWeightedRecallAndThreshold (const std::vector< float > &weights, float &recall, float &threshold) const
	Best weighted recall achievable with the corresponding threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedRecallPoint () const
	Best weighted recall achievable returned as a weighted precision/recall point.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedRecallPoint (const std::vector< float > &weights) const
	Best weighted recall achievable returned as a weighted precision/recall point.
float	GetPrecision () const
	Precision (proportion of detected objects that are correct).
float	GetWeightedPrecision () const
	Weighted average of the precision (proportion of detected objects that are correct) for each label.
float	GetWeightedPrecision (const std::vector< float > &weights) const
	Weighted average of the precision (proportion of detected objects that are correct) for each label.
float	GetLabelPrecision (const std::string &label) const
	Precision for the given label.
float	GetLabelPrecision (int labelId) const
	Precision for the given label.
float	GetBestWeightedPrecision () const
	Best weighted precision achievable by changing the detection threshold.
float	GetBestWeightedPrecision (const std::vector< float > &weights) const
	Best weighted precision achievable by changing the detection threshold.
float	GetBestWeightedPrecisionThreshold () const
	Detection threshold that gives the best weighted precision (ELocatorMetrics::GetWeightedPrecision).
float	GetBestWeightedPrecisionThreshold (const std::vector< float > &weights) const
	Detection threshold that gives the best weighted precision (ELocatorMetrics::GetWeightedPrecision).
void	GetBestWeightedPrecisionAndThreshold (float &precision, float &threshold) const
	Best weighted precision achievable with the corresponding threshold.
void	GetBestWeightedPrecisionAndThreshold (const std::vector< float > &weights, float &precision, float &threshold) const
	Best weighted precision achievable with the corresponding threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedPrecisionPoint () const
	Best weighted precision achievable returned as a weighted precision/recall point.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedPrecisionPoint (const std::vector< float > &weights) const
	Best weighted precision achievable returned as a weighted precision/recall point.
float	GetFScore () const
	F-Score (harmonic mean of ELocatorMetrics::GetRecall and ELocatorMetrics::GetPrecision).
float	GetWeightedFScore () const
	Weighted average of the F-Score for each label.
float	GetWeightedFScore (const std::vector< float > &weights) const
	Weighted average of the F-Score for each label.
float	GetLabelFScore (const std::string &label) const
	F-Score for the given label.
float	GetLabelFScore (int labelId) const
	F-Score for the given label.
float	GetBestWeightedFScore () const
	Best weighted F-Score achievable by changing the detection threshold.
float	GetBestWeightedFScore (const std::vector< float > &weights) const
	Best weighted F-Score achievable by changing the detection threshold.
float	GetBestWeightedFScoreThreshold () const
	Detection threshold that gives the best weighted F-Score (ELocatorMetrics::GetWeightedFScore).
float	GetBestWeightedFScoreThreshold (const std::vector< float > &weights) const
	Detection threshold that gives the best weighted F-Score (ELocatorMetrics::GetWeightedFScore).
void	GetBestWeightedFScoreAndThreshold (float &fScore, float &threshold) const
	Best weighted F-Score achievable with the corresponding threshold.
void	GetBestWeightedFScoreAndThreshold (const std::vector< float > &weights, float &fScore, float &threshold) const
	Best weighted F-Score achievable with the corresponding threshold.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedFScorePoint () const
	Best weighted F-Score achievable returned as a weighted precision/recall point.
eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint	GetBestWeightedFScorePoint (const std::vector< float > &weights) const
	Best weighted F-Score achievable returned as a weighted precision/recall point.
uint32_t	GetNumDetectedObjects () const
	Number of detected objects.
uint32_t	GetNumDetectedObjects (const std::string &label) const
	Number of detected objects.
uint32_t	GetNumDetectedObjects (int labelId) const
	Number of detected objects.
uint32_t	GetNumCorrectlyDetectedObjects () const
	Number of correctly detected objects.
uint32_t	GetNumCorrectlyDetectedObjects (const std::string &label) const
	Number of correctly detected objects.
uint32_t	GetNumCorrectlyDetectedObjects (int labelId) const
	Number of correctly detected objects.
uint32_t	GetNumUndetectedObjects () const
	Number of undetected objects.
uint32_t	GetNumUndetectedObjects (const std::string &label) const
	Number of undetected objects.
uint32_t	GetNumUndetectedObjects (int labelId) const
	Number of undetected objects.
float	GetAveragePrecision () const
	Average Precision when matching prediction and ground truth with a minimum proximity of ELocatorBase::SetSameLabelMaxObjectProximity.
float	GetAveragePrecision50 () const
	Average precision when matching prediction and ground truth with a minimum "proximity" of 0.5.
float	GetLabelAveragePrecision (int labelIdx) const
	Average precision for detection of objects from the given label.
float	GetAverageProximity () const
	Average proximity of the predicted objects to the ground truth objects.
float	GetIntersectionOverUnion () const
	Same as ELocatorMetrics::GetAverageProximity.
float	GetLabelAverageProximity (int labelIdx) const
	Average proximity of the predicted objects of the given label to the ground truth objects.
float	GetLabelIntersectionOverUnion (int labelIdx) const
	Same as ELocatorMetrics::GetLabelAverageProximity.
void	SetDetectionThreshold (float threshold)
	Threshold for detected objects.
float	GetDetectionThreshold () const
	Threshold for detected objects.
float	GetImageAccuracy () const
	Image accuracy: the proportion of images that are correctly detected to contain or not objects, regardless of their labels.
uint32_t	GetNumCorrectlyPredictedImagesWithObjects () const
	Number of images containing objects that are correctly predicted as containing objects, regardless of the labels of the objects. This is the number of true positives at the image level.
uint32_t	GetNumCorrectlyPredictedImagesWithoutObjects () const
	Number of images without objects that are correctly predicted as containing no object. This is the number of true negatives at the image level.
uint32_t	GetNumBadlyPredictedImagesWithObjects () const
	Number of images with objects that are badly predicted as containing no object. This is the number of false negatives at the image level.
uint32_t	GetNumBadlyPredictedImagesWithoutObjects () const
	Number of images without objects that are badly predicted as containing objects. This is the number of false positives a the image level.
int	GetNumLabels () const
	Number of labels recognized by the ELocator tool that produced these metrics.
std::string	GetLabel (int idx) const
	Label.
	ELocatorMetrics (CppWrapperInternal::InternalConstructor)

Public Attributes
CppWrapperInternal::WrappedClass	wrappedClass_

Detailed Description

Collection of metrics used to evaluate the results of a ELocator tool.

License(s): BasicTypes

Constructor & Destructor Documentation

◆ ELocatorMetrics() [1/3]

eVision::EasyDeepLearning::ELocatorMetrics::ELocatorMetrics ( )

Constructs an ELocatorMetrics object.

◆ ELocatorMetrics() [2/3]

eVision::EasyDeepLearning::ELocatorMetrics::ELocatorMetrics ( const eVision::EasyDeepLearning::ELocatorMetrics & other )

Copy constructor of an ELocatorMetrics object.

Parameters

other Reference to the ELocatorMetrics object that should be copied

◆ ~ELocatorMetrics()

eVision::EasyDeepLearning::ELocatorMetrics::~ELocatorMetrics	(		)
throw	(	)

virtual

◆ ELocatorMetrics() [3/3]

eVision::EasyDeepLearning::ELocatorMetrics::ELocatorMetrics ( CppWrapperInternal::InternalConstructor )

Member Function Documentation

◆ GetAveragePrecision()

float eVision::EasyDeepLearning::ELocatorMetrics::GetAveragePrecision ( ) const

Average Precision when matching prediction and ground truth with a minimum proximity of ELocatorBase::SetSameLabelMaxObjectProximity.

◆ GetAveragePrecision50()

float eVision::EasyDeepLearning::ELocatorMetrics::GetAveragePrecision50 ( ) const

Average precision when matching prediction and ground truth with a minimum "proximity" of 0.5.

◆ GetAverageProximity()

float eVision::EasyDeepLearning::ELocatorMetrics::GetAverageProximity ( ) const

Average proximity of the predicted objects to the ground truth objects.

This value is the average of ELocatorMetrics::GetLabelAverageProximity over all the labels.

Look at the documentation of ELocator and EInterestPointLocator for their definition of proximity.

◆ GetBestWeightedFScore() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScore ( ) const

Best weighted F-Score achievable by changing the detection threshold.

◆ GetBestWeightedFScore() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScore ( const std::vector< float > & weights ) const

Best weighted F-Score achievable by changing the detection threshold.

Parameters

weights Label weights

◆ GetBestWeightedFScoreAndThreshold() [1/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScoreAndThreshold	(	const std::vector< float > &	weights,
		float &	fScore,
		float &	threshold ) const

Best weighted F-Score achievable with the corresponding threshold.

Parameters

weights	Label weights
fScore	Best weighted F-Score achievable
threshold	Threshold that achieves the best weighted F-Score

Deprecated

Use ELocatorMetrics::GetBestWeightedFScorePoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedFScoreAndThreshold() [2/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScoreAndThreshold	(	float &	fScore,
		float &	threshold ) const

Best weighted F-Score achievable with the corresponding threshold.

Parameters

fScore	Best weighted F-Score achievable
threshold	Threshold that achieves the best weighted F-Score

Deprecated

Use ELocatorMetrics::GetBestWeightedFScorePoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedFScorePoint() [1/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScorePoint ( ) const

Best weighted F-Score achievable returned as a weighted precision/recall point.

◆ GetBestWeightedFScorePoint() [2/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScorePoint ( const std::vector< float > & weights ) const

Best weighted F-Score achievable returned as a weighted precision/recall point.

Parameters

weights Label weights

◆ GetBestWeightedFScoreThreshold() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScoreThreshold ( ) const

Detection threshold that gives the best weighted F-Score (ELocatorMetrics::GetWeightedFScore).

◆ GetBestWeightedFScoreThreshold() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedFScoreThreshold ( const std::vector< float > & weights ) const

Detection threshold that gives the best weighted F-Score (ELocatorMetrics::GetWeightedFScore).

Parameters

weights Label weights

◆ GetBestWeightedPrecision() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecision ( ) const

Best weighted precision achievable by changing the detection threshold.

◆ GetBestWeightedPrecision() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecision ( const std::vector< float > & weights ) const

Best weighted precision achievable by changing the detection threshold.

Parameters

weights Label weights

◆ GetBestWeightedPrecisionAndThreshold() [1/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionAndThreshold	(	const std::vector< float > &	weights,
		float &	precision,
		float &	threshold ) const

Best weighted precision achievable with the corresponding threshold.

Parameters

weights	Label weights
precision	Best weighted precision achievable
threshold	Threshold that achieves the best weighted precision

Deprecated

Use ELocatorMetrics::GetBestWeightedPrecisionPoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedPrecisionAndThreshold() [2/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionAndThreshold	(	float &	precision,
		float &	threshold ) const

Best weighted precision achievable with the corresponding threshold.

Parameters

precision	Best weighted precision achievable
threshold	Threshold that achieves the best weighted precision

Deprecated

Use ELocatorMetrics::GetBestWeightedPrecisionPoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedPrecisionPoint() [1/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionPoint ( ) const

Best weighted precision achievable returned as a weighted precision/recall point.

◆ GetBestWeightedPrecisionPoint() [2/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionPoint ( const std::vector< float > & weights ) const

Best weighted precision achievable returned as a weighted precision/recall point.

Parameters

weights Label weights

◆ GetBestWeightedPrecisionThreshold() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionThreshold ( ) const

Detection threshold that gives the best weighted precision (ELocatorMetrics::GetWeightedPrecision).

◆ GetBestWeightedPrecisionThreshold() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedPrecisionThreshold ( const std::vector< float > & weights ) const

Detection threshold that gives the best weighted precision (ELocatorMetrics::GetWeightedPrecision).

Parameters

weights Label weights

◆ GetBestWeightedRecall() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecall ( ) const

Best weighted recall achievable by changing the detection threshold.

◆ GetBestWeightedRecall() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecall ( const std::vector< float > & weights ) const

Best weighted recall achievable by changing the detection threshold.

Parameters

weights Label weights

◆ GetBestWeightedRecallAndThreshold() [1/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallAndThreshold	(	const std::vector< float > &	weights,
		float &	recall,
		float &	threshold ) const

Best weighted recall achievable with the corresponding threshold.

Parameters

weights	Label weights
recall	Best weighted recall achievable
threshold	Threshold that achieves the best weighted recall

Deprecated

Use ELocatorMetrics::GetBestWeightedRecallPoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedRecallAndThreshold() [2/2]

void eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallAndThreshold	(	float &	recall,
		float &	threshold ) const

Best weighted recall achievable with the corresponding threshold.

Parameters

recall	Best weighted recall achievable
threshold	Threshold that achieves the best weighted recall

Deprecated

Use ELocatorMetrics::GetBestWeightedRecallPoint instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetBestWeightedRecallPoint() [1/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallPoint ( ) const

Best weighted recall achievable returned as a weighted precision/recall point.

◆ GetBestWeightedRecallPoint() [2/2]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallPoint ( const std::vector< float > & weights ) const

Best weighted recall achievable returned as a weighted precision/recall point.

Parameters

weights Label weights

◆ GetBestWeightedRecallThreshold() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallThreshold ( ) const

Detection threshold that gives the best weighted recall (ELocatorMetrics::GetWeightedRecall).

◆ GetBestWeightedRecallThreshold() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetBestWeightedRecallThreshold ( const std::vector< float > & weights ) const

Detection threshold that gives the best weighted recall (ELocatorMetrics::GetWeightedRecall).

Parameters

weights Label weights

◆ GetDetectionThreshold()

float eVision::EasyDeepLearning::ELocatorMetrics::GetDetectionThreshold ( ) const

Threshold for detected objects.

◆ GetError()

float eVision::EasyDeepLearning::ELocatorMetrics::GetError ( ) const

Error of the EasyLocate training algorithm (only available for the training metrics, see ELocatorBase::GetTrainingMetrics).

◆ GetFScore()

float eVision::EasyDeepLearning::ELocatorMetrics::GetFScore ( ) const

F-Score (harmonic mean of ELocatorMetrics::GetRecall and ELocatorMetrics::GetPrecision).

◆ GetImageAccuracy()

float eVision::EasyDeepLearning::ELocatorMetrics::GetImageAccuracy ( ) const

Image accuracy: the proportion of images that are correctly detected to contain or not objects, regardless of their labels.

◆ GetIntersectionOverUnion()

float eVision::EasyDeepLearning::ELocatorMetrics::GetIntersectionOverUnion ( ) const

Same as ELocatorMetrics::GetAverageProximity.

Deprecated

Use ELocatorMetrics::GetAverageProximity instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetLabel()

std::string eVision::EasyDeepLearning::ELocatorMetrics::GetLabel ( int idx ) const

Label.

Parameters

idx	Index of the label.

◆ GetLabelAveragePrecision()

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelAveragePrecision ( int labelIdx ) const

Average precision for detection of objects from the given label.

Parameters

labelIdx Label index

◆ GetLabelAverageProximity()

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelAverageProximity ( int labelIdx ) const

Average proximity of the predicted objects of the given label to the ground truth objects.

Parameters

labelIdx Index of the label.

Look at the documentation of ELocator and EInterestPointLocator for their definition of proximity.

◆ GetLabelFScore() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelFScore ( const std::string & label ) const

F-Score for the given label.

Parameters

label Label

◆ GetLabelFScore() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelFScore ( int labelId ) const

F-Score for the given label.

Parameters

labelId Label index

◆ GetLabelIntersectionOverUnion()

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelIntersectionOverUnion ( int labelIdx ) const

Same as ELocatorMetrics::GetLabelAverageProximity.

Parameters

labelIdx Index of the label.

Deprecated

Use ELocatorMetrics::GetLabelAverageProximity instead.

Deprecated in version 22.8. Scheduled to be removed in version 27.6.

◆ GetLabelPrecision() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelPrecision ( const std::string & label ) const

Precision for the given label.

Parameters

label Label

◆ GetLabelPrecision() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelPrecision ( int labelId ) const

Precision for the given label.

Parameters

labelId Label index

◆ GetLabelRecall() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelRecall ( const std::string & label ) const

Recall for the given label.

Parameters

label Label

◆ GetLabelRecall() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetLabelRecall ( int labelId ) const

Recall for the given label.

Parameters

labelId Label index

◆ GetNumBadlyPredictedImagesWithObjects()

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumBadlyPredictedImagesWithObjects ( ) const

Number of images with objects that are badly predicted as containing no object. This is the number of false negatives at the image level.

◆ GetNumBadlyPredictedImagesWithoutObjects()

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumBadlyPredictedImagesWithoutObjects ( ) const

Number of images without objects that are badly predicted as containing objects. This is the number of false positives a the image level.

◆ GetNumCorrectlyDetectedObjects() [1/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumCorrectlyDetectedObjects ( ) const

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than ELocatorBase::SetSameLabelMaxObjectProximity with a ground truth object of the same label.

A label can be specified to obtain the number of correctly detected objects for that label. Otherwise, the number of correctly detected objects is for all the labels.

◆ GetNumCorrectlyDetectedObjects() [2/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumCorrectlyDetectedObjects ( const std::string & label ) const

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than ELocatorBase::SetSameLabelMaxObjectProximity with a ground truth object of the same label.

A label can be specified to obtain the number of correctly detected objects for that label. Otherwise, the number of correctly detected objects is for all the labels.

Parameters

label Label

◆ GetNumCorrectlyDetectedObjects() [3/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumCorrectlyDetectedObjects ( int labelId ) const

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than ELocatorBase::SetSameLabelMaxObjectProximity with a ground truth object of the same label.

A label can be specified to obtain the number of correctly detected objects for that label. Otherwise, the number of correctly detected objects is for all the labels.

Parameters

labelId Index of the label

◆ GetNumCorrectlyPredictedImagesWithObjects()

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumCorrectlyPredictedImagesWithObjects ( ) const

Number of images containing objects that are correctly predicted as containing objects, regardless of the labels of the objects. This is the number of true positives at the image level.

◆ GetNumCorrectlyPredictedImagesWithoutObjects()

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumCorrectlyPredictedImagesWithoutObjects ( ) const

Number of images without objects that are correctly predicted as containing no object. This is the number of true negatives at the image level.

◆ GetNumDetectedObjects() [1/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumDetectedObjects ( ) const

Number of detected objects.

A label can be specified to obtain the number of detected objects for that label. Otherwise, the number of detected objects is for all the labels.

◆ GetNumDetectedObjects() [2/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumDetectedObjects ( const std::string & label ) const

Number of detected objects.

A label can be specified to obtain the number of detected objects for that label. Otherwise, the number of detected objects is for all the labels.

Parameters

label Label

◆ GetNumDetectedObjects() [3/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumDetectedObjects ( int labelId ) const

Number of detected objects.

A label can be specified to obtain the number of detected objects for that label. Otherwise, the number of detected objects is for all the labels.

Parameters

labelId Index of the label

◆ GetNumLabels()

int eVision::EasyDeepLearning::ELocatorMetrics::GetNumLabels ( ) const

Number of labels recognized by the ELocator tool that produced these metrics.

◆ GetNumUndetectedObjects() [1/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumUndetectedObjects ( ) const

Number of undetected objects.

An undetected object is a ground truth object that is not matched to any predicted object of the same label with a proximity higher than ELocatorBase::SetSameLabelMaxObjectProximity.

A label can be specified to obtain the number of undetected objects for that label. Otherwise, the number of undetected objects is for all the labels.

◆ GetNumUndetectedObjects() [2/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumUndetectedObjects ( const std::string & label ) const

Number of undetected objects.

An undetected object is a ground truth object that is not matched to any predicted object of the same label with a proximity higher than ELocatorBase::SetSameLabelMaxObjectProximity.

A label can be specified to obtain the number of undetected objects for that label. Otherwise, the number of undetected objects is for all the labels.

Parameters

label Label

◆ GetNumUndetectedObjects() [3/3]

uint32_t eVision::EasyDeepLearning::ELocatorMetrics::GetNumUndetectedObjects ( int labelId ) const

Number of undetected objects.

An undetected object is a ground truth object that is not matched to any predicted object of the same label with a proximity higher than ELocatorBase::SetSameLabelMaxObjectProximity.

A label can be specified to obtain the number of undetected objects for that label. Otherwise, the number of undetected objects is for all the labels.

Parameters

labelId Index of the label

◆ GetPrecision()

float eVision::EasyDeepLearning::ELocatorMetrics::GetPrecision ( ) const

Precision (proportion of detected objects that are correct).

◆ GetRecall()

float eVision::EasyDeepLearning::ELocatorMetrics::GetRecall ( ) const

Recall (true positive rate, proportion of ground truth object that are correctly detected).

◆ GetWeightedFScore() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedFScore ( ) const

Weighted average of the F-Score for each label.

◆ GetWeightedFScore() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedFScore ( const std::vector< float > & weights ) const

Weighted average of the F-Score for each label.

Parameters

weights Label weights

◆ GetWeightedPrecision() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecision ( ) const

Weighted average of the precision (proportion of detected objects that are correct) for each label.

◆ GetWeightedPrecision() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecision ( const std::vector< float > & weights ) const

Weighted average of the precision (proportion of detected objects that are correct) for each label.

Parameters

weights Label weights

◆ GetWeightedPrecisionRecallPoint() [1/4]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecisionRecallPoint ( ) const

Weighted precision and recall metrics for the current or specified threshold.

◆ GetWeightedPrecisionRecallPoint() [2/4]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecisionRecallPoint ( const std::vector< float > & weights ) const

Weighted precision and recall metrics for the current or specified threshold.

Parameters

weights Label weights

◆ GetWeightedPrecisionRecallPoint() [3/4]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecisionRecallPoint ( float threshold ) const

Weighted precision and recall metrics for the current or specified threshold.

Parameters

threshold Threshold for which to get the weighted precision/recall point

◆ GetWeightedPrecisionRecallPoint() [4/4]

eVision::EasyDeepLearning::EWeightedPrecisionRecallPoint eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedPrecisionRecallPoint	(	float	threshold,
		const std::vector< float > &	weights ) const

Weighted precision and recall metrics for the current or specified threshold.

Parameters

threshold	Threshold for which to get the weighted precision/recall point
weights	Label weights

◆ GetWeightedRecall() [1/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedRecall ( ) const

Weighted average of the recall (true positive rate) for each label.

◆ GetWeightedRecall() [2/2]

float eVision::EasyDeepLearning::ELocatorMetrics::GetWeightedRecall ( const std::vector< float > & weights ) const

Weighted average of the recall (true positive rate) for each label.

Parameters

weights Label weights

◆ IsValid()

bool eVision::EasyDeepLearning::ELocatorMetrics::IsValid ( ) const

Indicates whether the object contains at least one result.

◆ Load() [1/2]

void eVision::EasyDeepLearning::ELocatorMetrics::Load ( const std::string & path )

Loads a locator metric. The given ESerializer must have been created for reading.

Parameters

path	The file path.

◆ Load() [2/2]

void eVision::EasyDeepLearning::ELocatorMetrics::Load ( eVision::ESerializer * serializer )

Loads a locator metric. The given ESerializer must have been created for reading.

Parameters

serializer The serializer.

◆ operator=()

eVision::EasyDeepLearning::ELocatorMetrics & eVision::EasyDeepLearning::ELocatorMetrics::operator= ( const eVision::EasyDeepLearning::ELocatorMetrics & other )

Assignment operator.

Parameters

other Reference to the ELocatorMetrics object used for the assignment

◆ Save() [1/2]

void eVision::EasyDeepLearning::ELocatorMetrics::Save ( const std::string & path )

Saves a locator metric. The given ESerializer must have been created for writing.

Parameters

path	The file path.

◆ Save() [2/2]

void eVision::EasyDeepLearning::ELocatorMetrics::Save ( eVision::ESerializer * serializer )

Saves a locator metric. The given ESerializer must have been created for writing.

Parameters

serializer The ESerializer object that is written to.

◆ SetDetectionThreshold()

void eVision::EasyDeepLearning::ELocatorMetrics::SetDetectionThreshold ( float threshold )

Threshold for detected objects.

Parameters

threshold Threshold

Member Data Documentation

◆ wrappedClass_

CppWrapperInternal::WrappedClass eVision::EasyDeepLearning::ELocatorMetrics::wrappedClass_

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

◆ ELocatorMetrics() [1/3]

◆ ELocatorMetrics() [2/3]

◆ ~ELocatorMetrics()

◆ ELocatorMetrics() [3/3]

Member Function Documentation

◆ GetAveragePrecision()

◆ GetAveragePrecision50()

◆ GetAverageProximity()

◆ GetBestWeightedFScore() [1/2]

◆ GetBestWeightedFScore() [2/2]

◆ GetBestWeightedFScoreAndThreshold() [1/2]

◆ GetBestWeightedFScoreAndThreshold() [2/2]

◆ GetBestWeightedFScorePoint() [1/2]

◆ GetBestWeightedFScorePoint() [2/2]

◆ GetBestWeightedFScoreThreshold() [1/2]

◆ GetBestWeightedFScoreThreshold() [2/2]

◆ GetBestWeightedPrecision() [1/2]

◆ GetBestWeightedPrecision() [2/2]

◆ GetBestWeightedPrecisionAndThreshold() [1/2]

◆ GetBestWeightedPrecisionAndThreshold() [2/2]

◆ GetBestWeightedPrecisionPoint() [1/2]

◆ GetBestWeightedPrecisionPoint() [2/2]

◆ GetBestWeightedPrecisionThreshold() [1/2]

◆ GetBestWeightedPrecisionThreshold() [2/2]

◆ GetBestWeightedRecall() [1/2]

◆ GetBestWeightedRecall() [2/2]

◆ GetBestWeightedRecallAndThreshold() [1/2]

◆ GetBestWeightedRecallAndThreshold() [2/2]

◆ GetBestWeightedRecallPoint() [1/2]

◆ GetBestWeightedRecallPoint() [2/2]

◆ GetBestWeightedRecallThreshold() [1/2]

◆ GetBestWeightedRecallThreshold() [2/2]

◆ GetDetectionThreshold()

◆ GetError()

◆ GetFScore()

◆ GetImageAccuracy()

◆ GetIntersectionOverUnion()

◆ GetLabel()

◆ GetLabelAveragePrecision()

◆ GetLabelAverageProximity()

◆ GetLabelFScore() [1/2]

◆ GetLabelFScore() [2/2]

◆ GetLabelIntersectionOverUnion()

◆ GetLabelPrecision() [1/2]

◆ GetLabelPrecision() [2/2]

◆ GetLabelRecall() [1/2]

◆ GetLabelRecall() [2/2]

◆ GetNumBadlyPredictedImagesWithObjects()

◆ GetNumBadlyPredictedImagesWithoutObjects()

◆ GetNumCorrectlyDetectedObjects() [1/3]

◆ GetNumCorrectlyDetectedObjects() [2/3]

◆ GetNumCorrectlyDetectedObjects() [3/3]

◆ GetNumCorrectlyPredictedImagesWithObjects()

◆ GetNumCorrectlyPredictedImagesWithoutObjects()

◆ GetNumDetectedObjects() [1/3]

◆ GetNumDetectedObjects() [2/3]

◆ GetNumDetectedObjects() [3/3]

◆ GetNumLabels()

◆ GetNumUndetectedObjects() [1/3]

◆ GetNumUndetectedObjects() [2/3]

◆ GetNumUndetectedObjects() [3/3]

◆ GetPrecision()

◆ GetRecall()

◆ GetWeightedFScore() [1/2]

◆ GetWeightedFScore() [2/2]

◆ GetWeightedPrecision() [1/2]

◆ GetWeightedPrecision() [2/2]

◆ GetWeightedPrecisionRecallPoint() [1/4]

◆ GetWeightedPrecisionRecallPoint() [2/4]

◆ GetWeightedPrecisionRecallPoint() [3/4]

◆ GetWeightedPrecisionRecallPoint() [4/4]

◆ GetWeightedRecall() [1/2]

◆ GetWeightedRecall() [2/2]

◆ IsValid()

◆ Load() [1/2]

◆ Load() [2/2]