eVision Python API
Loading...
Searching...
No Matches
eVision.EasyDeepLearning.ELocatorMetrics Class Reference

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

Public Member Functions

None __init__ (self)
 Constructs an eVision.EasyDeepLearning.ELocatorMetrics object.
None __init__ (self, eVision.EasyDeepLearning.ELocatorMetrics other)
 Copy constructor of an eVision.EasyDeepLearning.ELocatorMetrics object.
None Load (self, str path)
 Loads a locator metric.
None Load (self, eVision.ESerializer|None serializer)
 Loads a locator metric.
None Save (self, str path)
 Saves a locator metric.
None Save (self, eVision.ESerializer|None serializer)
 Saves a locator metric.
bool IsValid (self)
 Indicates whether the object contains at least one result.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetWeightedPrecisionRecallPoint (self)
 Weighted precision and recall metrics for the current or specified threshold.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetWeightedPrecisionRecallPoint (self, float threshold)
 Weighted precision and recall metrics for the current or specified threshold.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetWeightedPrecisionRecallPoint (self, list[float] weights)
 Weighted precision and recall metrics for the current or specified threshold.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetWeightedPrecisionRecallPoint (self, float threshold, list[float] weights)
 Weighted precision and recall metrics for the current or specified threshold.
float GetWeightedRecall (self)
 Weighted average of the recall (true positive rate) for each label.
float GetWeightedRecall (self, list[float] weights)
 Weighted average of the recall (true positive rate) for each label.
float GetLabelRecall (self, str label)
 Recall for the given label.
float GetLabelRecall (self, int labelId)
 Recall for the given label.
float GetBestWeightedRecall (self)
 Best weighted recall achievable by changing the detection threshold.
float GetBestWeightedRecall (self, list[float] weights)
 Best weighted recall achievable by changing the detection threshold.
float GetBestWeightedRecallThreshold (self)
 Detection threshold that gives the best weighted recall (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedRecall).
float GetBestWeightedRecallThreshold (self, list[float] weights)
 Detection threshold that gives the best weighted recall (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedRecall).
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedRecallPoint (self)
 Best weighted recall achievable returned as a weighted precision/recall point.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedRecallPoint (self, list[float] weights)
 Best weighted recall achievable returned as a weighted precision/recall point.
float GetWeightedPrecision (self)
 Weighted average of the precision (proportion of detected objects that are correct) for each label.
float GetWeightedPrecision (self, list[float] weights)
 Weighted average of the precision (proportion of detected objects that are correct) for each label.
float GetLabelPrecision (self, str label)
 Precision for the given label.
float GetLabelPrecision (self, int labelId)
 Precision for the given label.
float GetBestWeightedPrecision (self)
 Best weighted precision achievable by changing the detection threshold.
float GetBestWeightedPrecision (self, list[float] weights)
 Best weighted precision achievable by changing the detection threshold.
float GetBestWeightedPrecisionThreshold (self)
 Detection threshold that gives the best weighted precision (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecision).
float GetBestWeightedPrecisionThreshold (self, list[float] weights)
 Detection threshold that gives the best weighted precision (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecision).
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedPrecisionPoint (self)
 Best weighted precision achievable returned as a weighted precision/recall point.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedPrecisionPoint (self, list[float] weights)
 Best weighted precision achievable returned as a weighted precision/recall point.
float GetWeightedFScore (self)
 Weighted average of the F-Score for each label.
float GetWeightedFScore (self, list[float] weights)
 Weighted average of the F-Score for each label.
float GetLabelFScore (self, str label)
 F-Score for the given label.
float GetLabelFScore (self, int labelId)
 F-Score for the given label.
float GetBestWeightedFScore (self)
 Best weighted F-Score achievable by changing the detection threshold.
float GetBestWeightedFScore (self, list[float] weights)
 Best weighted F-Score achievable by changing the detection threshold.
float GetBestWeightedFScoreThreshold (self)
 Detection threshold that gives the best weighted F-Score (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedFScore).
float GetBestWeightedFScoreThreshold (self, list[float] weights)
 Detection threshold that gives the best weighted F-Score (eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedFScore).
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedFScorePoint (self)
 Best weighted F-Score achievable returned as a weighted precision/recall point.
eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint GetBestWeightedFScorePoint (self, list[float] weights)
 Best weighted F-Score achievable returned as a weighted precision/recall point.
int GetNumDetectedObjects (self)
 Number of detected objects.
int GetNumDetectedObjects (self, str label)
 Number of detected objects.
int GetNumDetectedObjects (self, int labelId)
 Number of detected objects.
int GetNumCorrectlyDetectedObjects (self)
 Number of correctly detected objects.
int GetNumCorrectlyDetectedObjects (self, str label)
 Number of correctly detected objects.
int GetNumCorrectlyDetectedObjects (self, int labelId)
 Number of correctly detected objects.
int GetNumUndetectedObjects (self)
 Number of undetected objects.
int GetNumUndetectedObjects (self, str label)
 Number of undetected objects.
int GetNumUndetectedObjects (self, int labelId)
 Number of undetected objects.
float GetLabelAveragePrecision (self, int labelIdx)
 Average precision for detection of objects from the given label.
float GetLabelAverageProximity (self, int labelIdx)
 Average proximity of the predicted objects of the given label to the ground truth objects.
float GetLabelIntersectionOverUnion (self, int labelIdx)
 Same as eVision.EasyDeepLearning.ELocatorMetrics.GetLabelAverageProximity.
str GetLabel (self, int idx)
 Label.

Protected Member Functions

float _Error__getter (self)
 \cond
float _Recall__getter (self)
 \cond
float _Precision__getter (self)
 \cond
float _FScore__getter (self)
 \cond
float _AveragePrecision__getter (self)
 \cond
float _AveragePrecision50__getter (self)
 \cond
float _AverageProximity__getter (self)
 \cond
float _IntersectionOverUnion__getter (self)
 \cond
float _DetectionThreshold__getter (self)
 \cond
 _DetectionThreshold__setter (self, float value)
 Threshold for detected objects.
float _ImageAccuracy__getter (self)
 \cond
int _NumCorrectlyPredictedImagesWithObjects__getter (self)
 \cond
int _NumCorrectlyPredictedImagesWithoutObjects__getter (self)
 \cond
int _NumBadlyPredictedImagesWithObjects__getter (self)
 \cond
int _NumBadlyPredictedImagesWithoutObjects__getter (self)
 \cond
int _NumLabels__getter (self)
 \cond

Properties

 Error = property(_Error__getter, None)
 \endcond
 Recall = property(_Recall__getter, None)
 \endcond
 Precision = property(_Precision__getter, None)
 \endcond
 FScore = property(_FScore__getter, None)
 \endcond
 AveragePrecision = property(_AveragePrecision__getter, None)
 \endcond
 AveragePrecision50 = property(_AveragePrecision50__getter, None)
 \endcond
 AverageProximity = property(_AverageProximity__getter, None)
 \endcond
 IntersectionOverUnion = property(_IntersectionOverUnion__getter, None)
 \endcond
 DetectionThreshold = property(_DetectionThreshold__getter, _DetectionThreshold__setter)
 \endcond
 ImageAccuracy = property(_ImageAccuracy__getter, None)
 \endcond
 NumCorrectlyPredictedImagesWithObjects = property(_NumCorrectlyPredictedImagesWithObjects__getter, None)
 \endcond
 NumCorrectlyPredictedImagesWithoutObjects = property(_NumCorrectlyPredictedImagesWithoutObjects__getter, None)
 \endcond
 NumBadlyPredictedImagesWithObjects = property(_NumBadlyPredictedImagesWithObjects__getter, None)
 \endcond
 NumBadlyPredictedImagesWithoutObjects = property(_NumBadlyPredictedImagesWithoutObjects__getter, None)
 \endcond
 NumLabels = property(_NumLabels__getter, None)
 \endcond

Detailed Description

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

Remarks:

License(s): BasicTypes

Constructor & Destructor Documentation

◆ __init__() [1/2]

None eVision.EasyDeepLearning.ELocatorMetrics.__init__ ( self)

Constructs an eVision.EasyDeepLearning.ELocatorMetrics object.

◆ __init__() [2/2]

None eVision.EasyDeepLearning.ELocatorMetrics.__init__ ( self,
eVision.EasyDeepLearning.ELocatorMetrics other )

Copy constructor of an eVision.EasyDeepLearning.ELocatorMetrics object.

Parameters
otherReference to the eVision.EasyDeepLearning.ELocatorMetrics object that should be copied

Member Function Documentation

◆ _AveragePrecision50__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._AveragePrecision50__getter ( self)
protected

\cond

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

◆ _AveragePrecision__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._AveragePrecision__getter ( self)
protected

\cond

Average Precision when matching prediction and ground truth with a minimum proximity of eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity.

◆ _AverageProximity__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._AverageProximity__getter ( self)
protected

\cond

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

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

Remarks:

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

◆ _DetectionThreshold__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._DetectionThreshold__getter ( self)
protected

\cond

Threshold for detected objects.

◆ _DetectionThreshold__setter()

eVision.EasyDeepLearning.ELocatorMetrics._DetectionThreshold__setter ( self,
float value )
protected

Threshold for detected objects.

◆ _Error__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._Error__getter ( self)
protected

\cond

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

◆ _FScore__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._FScore__getter ( self)
protected

\cond

F-Score (harmonic mean of eVision.EasyDeepLearning.ELocatorMetrics.Recall and eVision.EasyDeepLearning.ELocatorMetrics.Precision).

◆ _ImageAccuracy__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._ImageAccuracy__getter ( self)
protected

\cond

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

◆ _IntersectionOverUnion__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._IntersectionOverUnion__getter ( self)
protected

\cond

Same as eVision.EasyDeepLearning.ELocatorMetrics.AverageProximity.

Deprecated

Use eVision.EasyDeepLearning.ELocatorMetrics.AverageProximity instead.

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

◆ _NumBadlyPredictedImagesWithObjects__getter()

int eVision.EasyDeepLearning.ELocatorMetrics._NumBadlyPredictedImagesWithObjects__getter ( self)
protected

\cond

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

◆ _NumBadlyPredictedImagesWithoutObjects__getter()

int eVision.EasyDeepLearning.ELocatorMetrics._NumBadlyPredictedImagesWithoutObjects__getter ( self)
protected

\cond

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

◆ _NumCorrectlyPredictedImagesWithObjects__getter()

int eVision.EasyDeepLearning.ELocatorMetrics._NumCorrectlyPredictedImagesWithObjects__getter ( self)
protected

\cond

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.

◆ _NumCorrectlyPredictedImagesWithoutObjects__getter()

int eVision.EasyDeepLearning.ELocatorMetrics._NumCorrectlyPredictedImagesWithoutObjects__getter ( self)
protected

\cond

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

◆ _NumLabels__getter()

int eVision.EasyDeepLearning.ELocatorMetrics._NumLabels__getter ( self)
protected

\cond

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

◆ _Precision__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._Precision__getter ( self)
protected

\cond

Precision (proportion of detected objects that are correct).

◆ _Recall__getter()

float eVision.EasyDeepLearning.ELocatorMetrics._Recall__getter ( self)
protected

\cond

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

◆ GetBestWeightedFScore() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScore ( self)

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

◆ GetBestWeightedFScore() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScore ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedFScorePoint() [1/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScorePoint ( self)

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

◆ GetBestWeightedFScorePoint() [2/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScorePoint ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedFScoreThreshold() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScoreThreshold ( self)

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

◆ GetBestWeightedFScoreThreshold() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedFScoreThreshold ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedPrecision() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecision ( self)

Best weighted precision achievable by changing the detection threshold.

◆ GetBestWeightedPrecision() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecision ( self,
list[float] weights )

Best weighted precision achievable by changing the detection threshold.

Parameters
weightsLabel weights

◆ GetBestWeightedPrecisionPoint() [1/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecisionPoint ( self)

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

◆ GetBestWeightedPrecisionPoint() [2/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecisionPoint ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedPrecisionThreshold() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecisionThreshold ( self)

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

◆ GetBestWeightedPrecisionThreshold() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedPrecisionThreshold ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedRecall() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecall ( self)

Best weighted recall achievable by changing the detection threshold.

◆ GetBestWeightedRecall() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecall ( self,
list[float] weights )

Best weighted recall achievable by changing the detection threshold.

Parameters
weightsLabel weights

◆ GetBestWeightedRecallPoint() [1/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecallPoint ( self)

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

◆ GetBestWeightedRecallPoint() [2/2]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecallPoint ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetBestWeightedRecallThreshold() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecallThreshold ( self)

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

◆ GetBestWeightedRecallThreshold() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetBestWeightedRecallThreshold ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetLabel()

str eVision.EasyDeepLearning.ELocatorMetrics.GetLabel ( self,
int idx )

Label.

Parameters
idxIndex of the label.

◆ GetLabelAveragePrecision()

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelAveragePrecision ( self,
int labelIdx )

Average precision for detection of objects from the given label.

Parameters
labelIdxLabel index

◆ GetLabelAverageProximity()

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelAverageProximity ( self,
int labelIdx )

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

Remarks:

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

Parameters
labelIdxIndex of the label.

◆ GetLabelFScore() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelFScore ( self,
int labelId )

F-Score for the given label.

Parameters
labelIdLabel index

◆ GetLabelFScore() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelFScore ( self,
str label )

F-Score for the given label.

Parameters
labelLabel

◆ GetLabelIntersectionOverUnion()

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelIntersectionOverUnion ( self,
int labelIdx )

Same as eVision.EasyDeepLearning.ELocatorMetrics.GetLabelAverageProximity.

Deprecated

Use eVision.EasyDeepLearning.ELocatorMetrics.GetLabelAverageProximity instead.

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

Parameters
labelIdxIndex of the label.

◆ GetLabelPrecision() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelPrecision ( self,
int labelId )

Precision for the given label.

Parameters
labelIdLabel index

◆ GetLabelPrecision() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelPrecision ( self,
str label )

Precision for the given label.

Parameters
labelLabel

◆ GetLabelRecall() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelRecall ( self,
int labelId )

Recall for the given label.

Parameters
labelIdLabel index

◆ GetLabelRecall() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetLabelRecall ( self,
str label )

Recall for the given label.

Parameters
labelLabel

◆ GetNumCorrectlyDetectedObjects() [1/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumCorrectlyDetectedObjects ( self)

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity 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]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumCorrectlyDetectedObjects ( self,
int labelId )

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity 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
labelIdIndex of the label

◆ GetNumCorrectlyDetectedObjects() [3/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumCorrectlyDetectedObjects ( self,
str label )

Number of correctly detected objects.

A correctly detected object is a predicted object that has a "proximity" higher than eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity 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
labelLabel

◆ GetNumDetectedObjects() [1/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumDetectedObjects ( self)

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]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumDetectedObjects ( self,
int labelId )

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
labelIdIndex of the label

◆ GetNumDetectedObjects() [3/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumDetectedObjects ( self,
str label )

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
labelLabel

◆ GetNumUndetectedObjects() [1/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumUndetectedObjects ( self)

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 eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity.

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]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumUndetectedObjects ( self,
int labelId )

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 eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity.

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
labelIdIndex of the label

◆ GetNumUndetectedObjects() [3/3]

int eVision.EasyDeepLearning.ELocatorMetrics.GetNumUndetectedObjects ( self,
str label )

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 eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity.

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
labelLabel

◆ GetWeightedFScore() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedFScore ( self)

Weighted average of the F-Score for each label.

◆ GetWeightedFScore() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedFScore ( self,
list[float] weights )

Weighted average of the F-Score for each label.

Parameters
weightsLabel weights

◆ GetWeightedPrecision() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecision ( self)

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

◆ GetWeightedPrecision() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecision ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetWeightedPrecisionRecallPoint() [1/4]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecisionRecallPoint ( self)

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

◆ GetWeightedPrecisionRecallPoint() [2/4]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecisionRecallPoint ( self,
float threshold )

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

Parameters
thresholdThreshold for which to get the weighted precision/recall point

◆ GetWeightedPrecisionRecallPoint() [3/4]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecisionRecallPoint ( self,
float threshold,
list[float] weights )

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

Parameters
thresholdThreshold for which to get the weighted precision/recall point
weightsLabel weights

◆ GetWeightedPrecisionRecallPoint() [4/4]

eVision.EasyDeepLearning.EWeightedPrecisionRecallPoint eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedPrecisionRecallPoint ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ GetWeightedRecall() [1/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedRecall ( self)

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

◆ GetWeightedRecall() [2/2]

float eVision.EasyDeepLearning.ELocatorMetrics.GetWeightedRecall ( self,
list[float] weights )

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

Parameters
weightsLabel weights

◆ IsValid()

bool eVision.EasyDeepLearning.ELocatorMetrics.IsValid ( self)

Indicates whether the object contains at least one result.

◆ Load() [1/2]

None eVision.EasyDeepLearning.ELocatorMetrics.Load ( self,
eVision.ESerializer | None serializer )

Loads a locator metric.

The given eVision.ESerializer must have been created for reading.

Parameters
serializerThe serializer.

◆ Load() [2/2]

None eVision.EasyDeepLearning.ELocatorMetrics.Load ( self,
str path )

Loads a locator metric.

The given eVision.ESerializer must have been created for reading.

Parameters
pathThe file path.

◆ Save() [1/2]

None eVision.EasyDeepLearning.ELocatorMetrics.Save ( self,
eVision.ESerializer | None serializer )

Saves a locator metric.

The given eVision.ESerializer must have been created for writing.

Parameters
serializerThe eVision.ESerializer object that is written to.

◆ Save() [2/2]

None eVision.EasyDeepLearning.ELocatorMetrics.Save ( self,
str path )

Saves a locator metric.

The given eVision.ESerializer must have been created for writing.

Parameters
pathThe file path.

Property Documentation

◆ AveragePrecision

eVision.EasyDeepLearning.ELocatorMetrics.AveragePrecision = property(_AveragePrecision__getter, None)
static

\endcond

Average Precision when matching prediction and ground truth with a minimum proximity of eVision.EasyDeepLearning.ELocatorBase.SameLabelMaxObjectProximity.

◆ AveragePrecision50

eVision.EasyDeepLearning.ELocatorMetrics.AveragePrecision50 = property(_AveragePrecision50__getter, None)
static

\endcond

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

◆ AverageProximity

eVision.EasyDeepLearning.ELocatorMetrics.AverageProximity = property(_AverageProximity__getter, None)
static

\endcond

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

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

Remarks:

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

◆ DetectionThreshold

eVision.EasyDeepLearning.ELocatorMetrics.DetectionThreshold = property(_DetectionThreshold__getter, _DetectionThreshold__setter)
static

\endcond

Threshold for detected objects.

◆ Error

eVision.EasyDeepLearning.ELocatorMetrics.Error = property(_Error__getter, None)
static

\endcond

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

◆ FScore

eVision.EasyDeepLearning.ELocatorMetrics.FScore = property(_FScore__getter, None)
static

\endcond

F-Score (harmonic mean of eVision.EasyDeepLearning.ELocatorMetrics.Recall and eVision.EasyDeepLearning.ELocatorMetrics.Precision).

◆ ImageAccuracy

eVision.EasyDeepLearning.ELocatorMetrics.ImageAccuracy = property(_ImageAccuracy__getter, None)
static

\endcond

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

◆ IntersectionOverUnion

eVision.EasyDeepLearning.ELocatorMetrics.IntersectionOverUnion = property(_IntersectionOverUnion__getter, None)
static

\endcond

Same as eVision.EasyDeepLearning.ELocatorMetrics.AverageProximity.

Deprecated

Use eVision.EasyDeepLearning.ELocatorMetrics.AverageProximity instead.

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

◆ NumBadlyPredictedImagesWithObjects

eVision.EasyDeepLearning.ELocatorMetrics.NumBadlyPredictedImagesWithObjects = property(_NumBadlyPredictedImagesWithObjects__getter, None)
static

\endcond

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

◆ NumBadlyPredictedImagesWithoutObjects

eVision.EasyDeepLearning.ELocatorMetrics.NumBadlyPredictedImagesWithoutObjects = property(_NumBadlyPredictedImagesWithoutObjects__getter, None)
static

\endcond

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

◆ NumCorrectlyPredictedImagesWithObjects

eVision.EasyDeepLearning.ELocatorMetrics.NumCorrectlyPredictedImagesWithObjects = property(_NumCorrectlyPredictedImagesWithObjects__getter, None)
static

\endcond

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.

◆ NumCorrectlyPredictedImagesWithoutObjects

eVision.EasyDeepLearning.ELocatorMetrics.NumCorrectlyPredictedImagesWithoutObjects = property(_NumCorrectlyPredictedImagesWithoutObjects__getter, None)
static

\endcond

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

◆ NumLabels

eVision.EasyDeepLearning.ELocatorMetrics.NumLabels = property(_NumLabels__getter, None)
static

\endcond

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

◆ Precision

eVision.EasyDeepLearning.ELocatorMetrics.Precision = property(_Precision__getter, None)
static

\endcond

Precision (proportion of detected objects that are correct).

◆ Recall

eVision.EasyDeepLearning.ELocatorMetrics.Recall = property(_Recall__getter, None)
static

\endcond

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