It provides quality measures as defined in ISO/IEC 29794-5:2025. More...

#include <OFIQFace.h>

Collaboration diagram for OFIQFace:

Public Member Functions
void	WriteTo (Writer &writer) const final
	Serialize OFIQFace to Writer.
Rectangle	GetBoundingBox () const
	Provides bounding box of detected face.
uint8_t	GetSharpness () const
	Provides sharpness of the face.
uint8_t	GetUnifiedQualityScore () const
	Unified quality score (QS) mapping native OFIQ quality.
uint8_t	GetBackgroundUniformity () const
	Background uniformity of the face image.
uint8_t	GetIlluminationUniformity () const
	Illumination uniformity across the face.
uint8_t	GetLuminanceMean () const
	Luminance mean of the face image.
uint8_t	GetLuminanceVariance () const
	Luminance variance in the face image.
uint8_t	GetUnderExposurePrevention () const
	Under-exposure prevention.
uint8_t	GetOverExposurePrevention () const
	Over-exposure prevention.
uint8_t	GetDynamicRange () const
	Dynamic range of the face image.
uint8_t	GetCompressionArtifacts () const
	Absence of compression artefacts.
uint8_t	GetNaturalColour () const
	Natural colour of the face image.
uint8_t	GetSingleFacePresent () const
	Single face present in the image.
uint8_t	GetEyesOpen () const
	Eyes open.
uint8_t	GetMouthClosed () const
	Mouth closed.
uint8_t	GetEyesVisible () const
	Eyes visible.
uint8_t	GetMouthOcclusionPrevention () const
	Mouth occlusion prevention.
uint8_t	GetFaceOcclusionPrevention () const
	Face occlusion prevention.
uint8_t	GetInterEyeDistance () const
	Inter-eye distance quality.
uint8_t	GetHeadSize () const
	Head size in the image.
uint8_t	GetLeftwardCropOfTheFaceImage () const
	Leftward crop of the face image.
uint8_t	GetRightwardCropOfTheFaceImage () const
	Rightward crop of the face image.
uint8_t	GetMarginAboveOfTheFaceImage () const
	Margin above the face image.
uint8_t	GetMarginBelowOfTheFaceImage () const
	Margin below the face image.
uint8_t	GetHeadPoseYaw () const
	Head pose yaw (left-right rotation).
uint8_t	GetHeadPosePitch () const
	Head pose pitch (up-down tilt).
uint8_t	GetHeadPoseRoll () const
	Head pose roll (tilt about the nose axis).
uint8_t	GetExpressionNeutrality () const
	Expression neutrality.
uint8_t	GetNoHeadCoverings () const
	No head coverings.

Static Public Member Functions
static std::shared_ptr< OFIQFace >	Create (const std::shared_ptr< FaceCapture > &capture, const std::shared_ptr< FaceExecutor > &executor)
	Create OFIQ from FaceCapture using the given executor.
static std::shared_ptr< OFIQFace >	Create (const std::shared_ptr< FaceCapture > &capture)
	Create OFIQ from FaceCapture using default executor set in GlobalFaceExecutor.

Detailed Description

It provides quality measures as defined in ISO/IEC 29794-5:2025.

To calculate qualities we are using OFIQ-project.

Member Function Documentation

◆ Create() [1/2]

std::shared_ptr< OFIQFace > Create ( const std::shared_ptr< FaceCapture > & capture )

inlinestatic

Create OFIQ from FaceCapture using default executor set in GlobalFaceExecutor.

It creates this instance and calculates all qualities at once.

Required Solvers

Total size of solvers is 635 MB.

References ofiq/expression_neutrality/grimmer/hse_1_2_C_adaboost.yml, ofiq/expression_neutrality/hsemotion/enet_b0_8_best_vgaf_embed_zeroed.onnx, ofiq/expression_neutrality/hsemotion/enet_b2_8_embed_zeroed.onnx, ofiq/face_detection/ssd_facedetect.caffemodel, ofiq/face_detection/ssd_facedetect.prototxt.txt, ofiq/face_landmark_estimation/ADNet.onnx, ofiq/face_occlusion_segmentation/face_occlusion_segmentation_ort.onnx, ofiq/face_parsing/bisenet_400.onnx, ofiq/head_pose_estimation/mb1_120x120.onnx, ofiq/no_compression_artifacts/ssim_248_model.onnx, ofiq/sharpness/face_sharpness_rtree.xml, ofiq/unified_quality_score/magface_iresnet50_norm.onnx

Parameters

capture FaceCapture

Returns: OFIQFace instance with calculated qualities

Exceptions

NullArgumentException	when capture is null
EnrollmentException	when OFIQ calculation fails

References Create(), and GlobalFaceExecutor::Get().

Here is the call graph for this function:

◆ Create() [2/2]

std::shared_ptr< OFIQFace > Create	(	const std::shared_ptr< FaceCapture > &	capture,
		const std::shared_ptr< FaceExecutor > &	executor )

inlinestatic

Create OFIQ from FaceCapture using the given executor.

It creates this instance and calculates all qualities at once.

Required Solvers

Total size of solvers is 635 MB.

References ofiq/expression_neutrality/grimmer/hse_1_2_C_adaboost.yml, ofiq/expression_neutrality/hsemotion/enet_b0_8_best_vgaf_embed_zeroed.onnx, ofiq/expression_neutrality/hsemotion/enet_b2_8_embed_zeroed.onnx, ofiq/face_detection/ssd_facedetect.caffemodel, ofiq/face_detection/ssd_facedetect.prototxt.txt, ofiq/face_landmark_estimation/ADNet.onnx, ofiq/face_occlusion_segmentation/face_occlusion_segmentation_ort.onnx, ofiq/face_parsing/bisenet_400.onnx, ofiq/head_pose_estimation/mb1_120x120.onnx, ofiq/no_compression_artifacts/ssim_248_model.onnx, ofiq/sharpness/face_sharpness_rtree.xml, ofiq/unified_quality_score/magface_iresnet50_norm.onnx

Parameters

capture	FaceCapture
executor	FaceExecutor used to execute calculation locally or on server

Returns: OFIQFace instance with calculated qualities

Exceptions

NullArgumentException	when capture is null
NullArgumentException	when executor is null
EnrollmentException	when OFIQ calculation fails

References GlobalLogger::Get().

Referenced by Create().

Here is the call graph for this function:

Here is the caller graph for this function:

◆ GetBackgroundUniformity()

uint8_t GetBackgroundUniformity ( ) const

inlinenodiscard

Background uniformity of the face image.

Measures how uniform the background is behind the face. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetBoundingBox()

Rectangle GetBoundingBox ( ) const

inlinenodiscard

Provides bounding box of detected face.

Returns: Bounding box of detected face

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetCompressionArtifacts()

uint8_t GetCompressionArtifacts ( ) const

inlinenodiscard

Absence of compression artefacts.

Measures degradation from lossy compression (e.g. JPEG). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetDynamicRange()

uint8_t GetDynamicRange ( ) const

inlinenodiscard

Dynamic range of the face image.

Reflects whether the image has sufficient range between darkest and brightest areas.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetExpressionNeutrality()

uint8_t GetExpressionNeutrality ( ) const

inlinenodiscard

Expression neutrality.

Indicates that the face has a neutral expression (e.g. mouth closed, no smile). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetEyesOpen()

uint8_t GetEyesOpen ( ) const

inlinenodiscard

Eyes open.

Indicates that both eyes are open (no closed or drooping eyelids). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetEyesVisible()

uint8_t GetEyesVisible ( ) const

inlinenodiscard

Eyes visible.

Indicates that both eyes are visible and not occluded.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetFaceOcclusionPrevention()

uint8_t GetFaceOcclusionPrevention ( ) const

inlinenodiscard

Face occlusion prevention.

Indicates that the face is not occluded by objects, hair, or other obstructions.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetHeadPosePitch()

uint8_t GetHeadPosePitch ( ) const

inlinenodiscard

Head pose pitch (up-down tilt).

Indicates vertical orientation of the head. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetHeadPoseRoll()

uint8_t GetHeadPoseRoll ( ) const

inlinenodiscard

Head pose roll (tilt about the nose axis).

Indicates whether the head is upright (no sideways tilt). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetHeadPoseYaw()

uint8_t GetHeadPoseYaw ( ) const

inlinenodiscard

Head pose yaw (left-right rotation).

Indicates how frontal the face is in the horizontal direction. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetHeadSize()

uint8_t GetHeadSize ( ) const

inlinenodiscard

Head size in the image.

Indicates whether the head size (e.g. face size) is within the expected range. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetIlluminationUniformity()

uint8_t GetIlluminationUniformity ( ) const

inlinenodiscard

Illumination uniformity across the face.

Measures uniformity of lighting on the face (e.g. absence of strong shadows or hotspots). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetInterEyeDistance()

uint8_t GetInterEyeDistance ( ) const

inlinenodiscard

Inter-eye distance quality.

Indicates whether the distance between the eyes is within the expected range for the image. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetLeftwardCropOfTheFaceImage()

uint8_t GetLeftwardCropOfTheFaceImage ( ) const

inlinenodiscard

Leftward crop of the face image.

Indicates sufficient margin to the left of the face in the frame.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetLuminanceMean()

uint8_t GetLuminanceMean ( ) const

inlinenodiscard

Luminance mean of the face image.

Indicates whether the face is neither under- nor over-exposed on average.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetLuminanceVariance()

uint8_t GetLuminanceVariance ( ) const

inlinenodiscard

Luminance variance in the face image.

Reflects variation in brightness across the face.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetMarginAboveOfTheFaceImage()

uint8_t GetMarginAboveOfTheFaceImage ( ) const

inlinenodiscard

Margin above the face image.

Indicates sufficient margin above the head in the frame. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetMarginBelowOfTheFaceImage()

uint8_t GetMarginBelowOfTheFaceImage ( ) const

inlinenodiscard

Margin below the face image.

Indicates sufficient margin below the chin in the frame. [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetMouthClosed()

uint8_t GetMouthClosed ( ) const

inlinenodiscard

Mouth closed.

Indicates that the mouth is closed (neutral expression). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetMouthOcclusionPrevention()

uint8_t GetMouthOcclusionPrevention ( ) const

inlinenodiscard

Mouth occlusion prevention.

Indicates that the mouth region is not occluded (e.g. by hands or objects).

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetNaturalColour()

uint8_t GetNaturalColour ( ) const

inlinenodiscard

Natural colour of the face image.

Indicates that skin and scene colours appear natural (no strong colour cast or enhancement). [8]

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetNoHeadCoverings()

uint8_t GetNoHeadCoverings ( ) const

inlinenodiscard

No head coverings.

Indicates that the face is not obscured by hats, caps, scarves, or similar coverings.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetOverExposurePrevention()

uint8_t GetOverExposurePrevention ( ) const

inlinenodiscard

Over-exposure prevention.

Indicates that the image is not too bright (over-exposed or washed out).

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetRightwardCropOfTheFaceImage()

uint8_t GetRightwardCropOfTheFaceImage ( ) const

inlinenodiscard

Rightward crop of the face image.

Indicates sufficient margin to the right of the face in the frame.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetSharpness()

uint8_t GetSharpness ( ) const

inlinenodiscard

Provides sharpness of the face.

Sharpness is the clarity of fine details in an image. It will be improved with precise focus, good imaging resolution, and absence of biometric capture subject motion relative to the camera. [8]

This quality component can be used to efficiently choose the better focused face portrait among several face samples of the same biometric capture subject. [8]

Returns: Sharpness score of the face in range [0, 100] where 100 is the best sharpness. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when sharpness calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetSingleFacePresent()

uint8_t GetSingleFacePresent ( ) const

inlinenodiscard

Single face present in the image.

Indicates that exactly one face is present and suitable for assessment.

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetUnderExposurePrevention()

uint8_t GetUnderExposurePrevention ( ) const

inlinenodiscard

Under-exposure prevention.

Indicates that the image is not too dark (under-exposed).

Returns: Score in range [0, 100], where 100 is best. Recommended decision threshold is around 95.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ GetUnifiedQualityScore()

uint8_t GetUnifiedQualityScore ( ) const

inlinenodiscard

Unified quality score (QS) mapping native OFIQ quality.

Predicts expected face recognition performance when compared to a canonical ISO/ICAO-compliant portrait.

Returns: Score in range [0, 100], where 100 indicates highest conformance. Recommended decision threshold is around 40.

Exceptions

EnrollmentException when the calculation fails

Referenced by WriteTo().

Here is the caller graph for this function:

◆ WriteTo()

void WriteTo ( Writer & writer ) const

inlinefinalvirtual

Serialize OFIQFace to Writer.

Parameters

writer Writer to serialize OFIQFace

Implements Serializable.

References GetBackgroundUniformity(), GetBoundingBox(), GetCompressionArtifacts(), GetDynamicRange(), GetExpressionNeutrality(), GetEyesOpen(), GetEyesVisible(), GetFaceOcclusionPrevention(), GetHeadPosePitch(), GetHeadPoseRoll(), GetHeadPoseYaw(), GetHeadSize(), GetIlluminationUniformity(), GetInterEyeDistance(), GetLeftwardCropOfTheFaceImage(), GetLuminanceMean(), GetLuminanceVariance(), GetMarginAboveOfTheFaceImage(), GetMarginBelowOfTheFaceImage(), GetMouthClosed(), GetMouthOcclusionPrevention(), GetNaturalColour(), GetNoHeadCoverings(), GetOverExposurePrevention(), GetRightwardCropOfTheFaceImage(), GetSharpness(), GetSingleFacePresent(), GetUnderExposurePrevention(), and GetUnifiedQualityScore().

Here is the call graph for this function:

The documentation for this class was generated from the following file:

Face/OFIQFace.h

Public Member Functions

Static Public Member Functions

Detailed Description

Member Function Documentation

◆ Create() [1/2]

◆ Create() [2/2]

◆ GetBackgroundUniformity()

◆ GetBoundingBox()

◆ GetCompressionArtifacts()

◆ GetDynamicRange()

◆ GetExpressionNeutrality()

◆ GetEyesOpen()

◆ GetEyesVisible()

◆ GetFaceOcclusionPrevention()

◆ GetHeadPosePitch()

◆ GetHeadPoseRoll()

◆ GetHeadPoseYaw()

◆ GetHeadSize()

◆ GetIlluminationUniformity()

◆ GetInterEyeDistance()

◆ GetLeftwardCropOfTheFaceImage()

◆ GetLuminanceMean()

◆ GetLuminanceVariance()

◆ GetMarginAboveOfTheFaceImage()

◆ GetMarginBelowOfTheFaceImage()

◆ GetMouthClosed()

◆ GetMouthOcclusionPrevention()

◆ GetNaturalColour()

◆ GetNoHeadCoverings()

◆ GetOverExposurePrevention()

◆ GetRightwardCropOfTheFaceImage()

◆ GetSharpness()

◆ GetSingleFacePresent()

◆ GetUnderExposurePrevention()

◆ GetUnifiedQualityScore()

◆ WriteTo()