An EDataAugmentation object is responsible for storing the data augmentation parameters and generating new transformed images from existing ones. Deep learning algorithms are not invariant to the location, scale, or rotation of the elements of interest in the image. Thus, if the final application requires the deep learning algorithm to be invariant to those characteristics, the dataset must contain images covering the spectrum of those characteristics. More...

#include <eVision/EasyDeepLearning/EDataAugmentation.h>

Collaboration diagram for eVision::EasyDeepLearning::EDataAugmentation:

Public Member Functions
	EDataAugmentation ()
	Creates an EDataAugmentation object.
	EDataAugmentation (const eVision::EasyDeepLearning::EDataAugmentation &other)
	Creates a copy of the EDataAugmentation object other.
virtual	~EDataAugmentation () throw ()
eVision::EasyDeepLearning::EDataAugmentation &	operator= (const eVision::EasyDeepLearning::EDataAugmentation &other)
	Copies the EDataAugmentation object other to this object.
bool	operator== (const eVision::EasyDeepLearning::EDataAugmentation &other) const
	Equality operator.
void	CopyTo (eVision::EasyDeepLearning::EDataAugmentation &other) const
	Copies itself to the EDataAugmentation object other.
void	Load (const std::string &path)
	Loads an EDataAugmentation object. The given ESerializer must have been created for reading.
void	Load (eVision::ESerializer *serializer)
	Loads an EDataAugmentation object. The given ESerializer must have been created for reading.
void	Save (const std::string &path)
	Saves an EDataAugmentation object. The given ESerializer must have been created for writing.
void	Save (eVision::ESerializer *serializer)
	Saves an EDataAugmentation object. The given ESerializer must have been created for writing.
bool	HasDataAugmentations () const
	Whether the EDataAugmentation object contains augmentations.
void	SetMaxHorizontalShift (int32_t maxShift)
	Sets the maximum horizontal shift allowed.
int32_t	GetMaxHorizontalShift () const
	Return the maximum horizontal shift allowed.
void	SetMaxVerticalShift (int32_t maxShift)
	Sets the maximum vertical shift allowed.
int32_t	GetMaxVerticalShift () const
	Return the maximum vertical shift allowed.
void	SetMaxRotationAngle (float maxAngle)
	Set the maximum rotation angle allowed. Its value must be between 0 and 180 degrees.
float	GetMaxRotationAngle () const
	Returns the maximum rotation angle allowed. The unit of the returned angle is in the current angle unit (see Easy::SetAngleUnit).
void	SetEnableVerticalFlip (bool enable)
	Sets whether to use vertically flipped versions of input images.
bool	GetEnableVerticalFlip () const
	Gets whether to use vertically flipped versions of input images.
void	SetEnableHorizontalFlip (bool enable)
	Sets whether to use horizontally flipped versions of input images.
bool	GetEnableHorizontalFlip () const
	Gets whether to use horizontally flipped versions of input images.
void	SetMinScale (float minScale)
	Sets the minimum scaling allowed. Its value must be strictly positive and below EDataAugmentation::SetMaxScale.
float	GetMinScale () const
	Gets the minimum scaling allowed.
void	SetMaxScale (float maxScale)
	Sets the maximum scaling allowed. Its value must be strictly positive and over EDataAugmentation::SetMinScale.
float	GetMaxScale () const
	Gets the maximum scaling allowed.
void	SetMaxHorizontalShear (float hShear)
	Sets the maximum absolute horizontal shear, represented as an angle from the vertical direction. Its value must be between 0 and 90 degrees.
float	GetMaxHorizontalShear () const
	Gets the maximum absolute horizontal shear, represented as an angle from the vertical direction.
void	SetMaxVerticalShear (float vShear)
	Sets the maximum absolute vertical shear, represented as an angle from the horizontal direction. Its value must be between 0 and 90 degrees.
float	GetMaxVerticalShear () const
	Gets the maximum absolute vertical shear, represented as an angle from the horizontal direction.
void	SetMaxBrightnessOffset (float offset)
	Sets the maximum absolute brightness offset. It must be between 0 and 1.
float	GetMaxBrightnessOffset () const
	Gets the maximum absolute brightness offset.
void	SetMaxContrastGain (float val)
	Sets the maximum contrast gain. Its value must be strictly positive and over EDataAugmentation::SetMinContrastGain.
float	GetMaxContrastGain () const
	Gets the maximum contrast gain.
void	SetMinContrastGain (float val)
	Sets the minimum contrast gain. Its value must be strictly positive and below EDataAugmentation::SetMaxContrastGain.
float	GetMinContrastGain () const
	Gets the minimum contrast gain.
void	SetMinGamma (float gamma)
	Sets the minimum gamma for gamma correction. Its value must be strictly positive and below EDataAugmentation::SetMaxGamma.
float	GetMinGamma () const
	Gets the minimum gamma for gamma correction.
void	SetMaxGamma (float gamma)
	Sets the maximum gamma for gamma correction. Its value must be higher than EDataAugmentation::SetMinGamma.
float	GetMaxGamma () const
	Gets the maximum gamma for gamma correction.
void	SetMaxHueOffset (float hueOffset)
	Sets the maximum absolute hue offset. Its value must be between 0 and 180 degrees.
float	GetMaxHueOffset () const
	Gets the maximum absolute hue offset.
void	SetMinSaturationGain (float saturationGain)
	Sets the minimum saturation gain. Its value must be strictly positive.
float	GetMinSaturationGain () const
	Gets the minimum saturation gain.
void	SetMaxSaturationGain (float saturationGain)
	Sets the maximum saturation gain. Its value must be over or equal to EDataAugmentation::SetMinSaturationGain.
float	GetMaxSaturationGain () const
	Gets the maximum saturation gain.
void	SetGaussianNoiseMaximumStandardDeviation (float gaussianMaximumDeviation)
	The Gaussian noise maximum standard deviation.
float	GetGaussianNoiseMaximumStandardDeviation () const
	The Gaussian noise maximum standard deviation.
void	SetGaussianNoiseMinimumStandardDeviation (float gaussianMinimumDeviation)
	The Gaussian noise minimum standard deviation.
float	GetGaussianNoiseMinimumStandardDeviation () const
	The Gaussian noise minimum standard deviation.
void	SetSpeckleNoiseMaximumStandardDeviation (float speckleNoiseMaximumStandardDeviation)
	The speckle noise maximum standard deviation.
float	GetSpeckleNoiseMaximumStandardDeviation () const
	The speckle noise maximum standard deviation.
void	SetSpeckleNoiseMinimumStandardDeviation (float speckleNoiseMinimumStandardDeviation)
	The speckle noise minimum standard deviation.
float	GetSpeckleNoiseMinimumStandardDeviation () const
	The speckle noise minimum standard deviation.
void	SetSaltAndPepperNoiseMaximumDensity (float saltAndPepperNoiseMaximumDensity)
	The maximum density of the salt and pepper noise.
float	GetSaltAndPepperNoiseMaximumDensity () const
	The maximum density of the salt and pepper noise.
void	SetSaltAndPepperNoiseMinimumDensity (float saltAndPepperNoiseMinimumDensity)
	The minimum density of the salt and pepper noise.
float	GetSaltAndPepperNoiseMinimumDensity () const
	The minimum density of the salt and pepper noise.
void	SetStainDisruptionMaxAnchorPoints (uint32_t stainDisruptioMaxAnchorPoints)
	The maximum number of anchor points to use while disrupting the ellipse.
uint32_t	GetStainDisruptionMaxAnchorPoints ()
	The maximum number of anchor points to use while disrupting the ellipse.
void	SetStainDisruptionMaxIntensity (float stainDisruptionMaxIntensity)
	The maximum offset applied to the coordinates (x,y) of a pixel from the ellipse contour.
float	GetStainDisruptionMaxIntensity ()
	The maximum offset applied to the coordinates (x,y) of a pixel from the ellipse contour.
void	SetStainEllipseMinRadius (uint32_t stainEllipseMinRadius)
	The minimum radius of the ellipse used to produce the stain.
uint32_t	GetStainEllipseMinRadius ()
	The minimum radius of the ellipse used to produce the stain.
void	SetStainEllipseMaxRadius (uint32_t stainEllipseMaxRadius)
	The maximum radius of the ellipse used to produce the stain.
uint32_t	GetStainEllipseMaxRadius ()
	The maximum radius of the ellipse used to produce the stain.
void	SetMinStainColor (uint8_t minStainColor)
	The minimum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).
uint8_t	GetMinStainColor ()
	The minimum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).
void	SetMaxStainColor (uint8_t maxStainColor)
	The maximum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).
uint8_t	GetMaxStainColor ()
	The maximum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).
void	SetStainColorVariation (uint32_t stainColorVariation)
	The variation of the color used to fill in the ellipse (We use an gaussian distribution of standard mean deviation stainColorVariation).
uint32_t	GetStainColorVariation ()
	The variation of the color used to fill in the ellipse (We use an gaussian distribution of standard mean deviation stainColorVariation).
void	SetStainInterpolationSiteFactor (float interpolationSiteFactor)
	The factor to compute the number of interpolation sites to use from the number of anchor points.
float	GetStainInterpolationSiteFactor ()
	The factor to compute the number of interpolation sites to use from the number of anchor points.
void	SetStainProbability (float stainProbability)
	Probability of applying the stain data augmentation (default value: 0, i.e. no stain).
float	GetStainProbability ()
	Probability of applying the stain data augmentation (default value: 0, i.e. no stain).
void	SetStainBlur (int blur)
	Stain blur.
int	GetStainBlur () const
	Stain blur.
std::unique_ptr< eVision::EBaseROI >	Generate (const eVision::EBaseROI *img, eVision::EasyDeepLearning::EDLDataAugmentationType type) const
	Generates a new image from a generic EBaseROI image. The caller is responsible for calling freeing the memory of the returned image by calling delete on the image.
std::unique_ptr< eVision::EBaseROI >	Generate (const eVision::EBaseROI *img) const
	Generates a new image from a generic EBaseROI image. The caller is responsible for calling freeing the memory of the returned image by calling delete on the image.
void	Generate (const eVision::EROIBW8 &source, eVision::EROIBW8 &dest, eVision::EasyDeepLearning::EDLDataAugmentationType type) const
	Generates a new image.
void	Generate (const eVision::EROIBW8 &source, eVision::EROIBW8 &dest) const
	Generates a new image.
void	Generate (const eVision::EROIBW16 &source, eVision::EROIBW16 &dest, eVision::EasyDeepLearning::EDLDataAugmentationType type) const
	Generates a new image.
void	Generate (const eVision::EROIBW16 &source, eVision::EROIBW16 &dest) const
	Generates a new image.
void	Generate (const eVision::EROIC24 &source, eVision::EROIC24 &dest, eVision::EasyDeepLearning::EDLDataAugmentationType type) const
	Generates a new image.
void	Generate (const eVision::EROIC24 &source, eVision::EROIC24 &dest) const
	Generates a new image.
	EDataAugmentation (CppWrapperInternal::InternalConstructor)

Public Attributes
CppWrapperInternal::WrappedClass	wrappedClass_

Detailed Description

An EDataAugmentation object is responsible for storing the data augmentation parameters and generating new transformed images from existing ones. Deep learning algorithms are not invariant to the location, scale, or rotation of the elements of interest in the image. Thus, if the final application requires the deep learning algorithm to be invariant to those characteristics, the dataset must contain images covering the spectrum of those characteristics.

Data augmentation can be used to avoid capturing in the original dataset the whole spectrum of those characteristics by automatically generating new versions of the images in the dataset that are shifted, scaled or rotated to cover this spectrum.

When generating a new image, the EDataAugmentation will randomly pick transformation value within the specified limits.

When using data augmentation, one must carefully chose the spectrum of transformations such that the element of interest are still visible in the image.

License(s): EasyClassify or EasySegment or EasyLocate

Constructor & Destructor Documentation

◆ EDataAugmentation() [1/3]

eVision::EasyDeepLearning::EDataAugmentation::EDataAugmentation ( )

Creates an EDataAugmentation object.

◆ EDataAugmentation() [2/3]

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

Creates a copy of the EDataAugmentation object other.

Parameters

other The other EDataAugmentation that we will copy into this one.

◆ ~EDataAugmentation()

eVision::EasyDeepLearning::EDataAugmentation::~EDataAugmentation	(		)
throw	(	)

virtual

◆ EDataAugmentation() [3/3]

eVision::EasyDeepLearning::EDataAugmentation::EDataAugmentation ( CppWrapperInternal::InternalConstructor )

Member Function Documentation

◆ CopyTo()

void eVision::EasyDeepLearning::EDataAugmentation::CopyTo ( eVision::EasyDeepLearning::EDataAugmentation & other ) const

Copies itself to the EDataAugmentation object other.

Parameters

other The other EDataAugmentation that we will copy into this one.

◆ Generate() [1/8]

std::unique_ptr< eVision::EBaseROI > eVision::EasyDeepLearning::EDataAugmentation::Generate ( const eVision::EBaseROI * img ) const

Generates a new image from a generic EBaseROI image. The caller is responsible for calling freeing the memory of the returned image by calling delete on the image.

Parameters

img	The image to transform.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [2/8]

std::unique_ptr< eVision::EBaseROI > eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EBaseROI *	img,
		eVision::EasyDeepLearning::EDLDataAugmentationType	type ) const

Generates a new image from a generic EBaseROI image. The caller is responsible for calling freeing the memory of the returned image by calling delete on the image.

Parameters

img	The image to transform.
type	The type of transformation to generate.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [3/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIBW16 &	source,
		eVision::EROIBW16 &	dest ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [4/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIBW16 &	source,
		eVision::EROIBW16 &	dest,
		eVision::EasyDeepLearning::EDLDataAugmentationType	type ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.
type	The type of transformation to generate.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [5/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIBW8 &	source,
		eVision::EROIBW8 &	dest ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [6/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIBW8 &	source,
		eVision::EROIBW8 &	dest,
		eVision::EasyDeepLearning::EDLDataAugmentationType	type ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.
type	The type of transformation to generate.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [7/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIC24 &	source,
		eVision::EROIC24 &	dest ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.

If the data augmentation fails, the method will throw an exception.

◆ Generate() [8/8]

void eVision::EasyDeepLearning::EDataAugmentation::Generate	(	const eVision::EROIC24 &	source,
		eVision::EROIC24 &	dest,
		eVision::EasyDeepLearning::EDLDataAugmentationType	type ) const

Generates a new image.

Parameters

source	The image to transform.
dest	The transformed image.
type	The type of transformation to generate.

If the data augmentation fails, the method will throw an exception.

◆ GetEnableHorizontalFlip()

bool eVision::EasyDeepLearning::EDataAugmentation::GetEnableHorizontalFlip ( ) const

Gets whether to use horizontally flipped versions of input images.

◆ GetEnableVerticalFlip()

bool eVision::EasyDeepLearning::EDataAugmentation::GetEnableVerticalFlip ( ) const

Gets whether to use vertically flipped versions of input images.

◆ GetGaussianNoiseMaximumStandardDeviation()

float eVision::EasyDeepLearning::EDataAugmentation::GetGaussianNoiseMaximumStandardDeviation ( ) const

The Gaussian noise maximum standard deviation.

The Gaussian noise is an additive noise sampled from a Gaussian distribution of deviation between EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.(also called the normal distribution).

Its value must be superior or equal to EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation.

This noise is computed before the salt and paper noise.

◆ GetGaussianNoiseMinimumStandardDeviation()

float eVision::EasyDeepLearning::EDataAugmentation::GetGaussianNoiseMinimumStandardDeviation ( ) const

The Gaussian noise minimum standard deviation.

The Gaussian noise is an additive noise sampled from a Gaussian distribution of deviation between EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.(also called the normal distribution).

Its value must be between 0 and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.

This noise is computed before the salt and paper noise.

◆ GetMaxBrightnessOffset()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxBrightnessOffset ( ) const

Gets the maximum absolute brightness offset.

◆ GetMaxContrastGain()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxContrastGain ( ) const

Gets the maximum contrast gain.

Contrast transformation is performed by multiplying each mean-centered pixel value by a gain value taken between EDataAugmentation::SetMinContrastGain and EDataAugmentation::SetMaxContrastGain. A contrast transformation does not change the overall brightness of an image.

◆ GetMaxGamma()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxGamma ( ) const

Gets the maximum gamma for gamma correction.

Gamma correction transformation is performed by raising each normalized pixel value to the power of gamma with gamma taken between EDataAugmentation::SetMinGamma and EDataAugmentation::SetMaxGamma.

◆ GetMaxHorizontalShear()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxHorizontalShear ( ) const

Gets the maximum absolute horizontal shear, represented as an angle from the vertical direction.

◆ GetMaxHorizontalShift()

int32_t eVision::EasyDeepLearning::EDataAugmentation::GetMaxHorizontalShift ( ) const

Return the maximum horizontal shift allowed.

◆ GetMaxHueOffset()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxHueOffset ( ) const

Gets the maximum absolute hue offset.

◆ GetMaxRotationAngle()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxRotationAngle ( ) const

Returns the maximum rotation angle allowed. The unit of the returned angle is in the current angle unit (see Easy::SetAngleUnit).

◆ GetMaxSaturationGain()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxSaturationGain ( ) const

Gets the maximum saturation gain.

◆ GetMaxScale()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxScale ( ) const

Gets the maximum scaling allowed.

◆ GetMaxStainColor()

uint8_t eVision::EasyDeepLearning::EDataAugmentation::GetMaxStainColor ( )

The maximum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).

Its value must be bigger than EDataAugmentation::SetMinStainColor.

◆ GetMaxVerticalShear()

float eVision::EasyDeepLearning::EDataAugmentation::GetMaxVerticalShear ( ) const

Gets the maximum absolute vertical shear, represented as an angle from the horizontal direction.

◆ GetMaxVerticalShift()

int32_t eVision::EasyDeepLearning::EDataAugmentation::GetMaxVerticalShift ( ) const

Return the maximum vertical shift allowed.

◆ GetMinContrastGain()

float eVision::EasyDeepLearning::EDataAugmentation::GetMinContrastGain ( ) const

Gets the minimum contrast gain.

Contrast transformation is performed by multiplying each mean-centered pixel value by a gain value taken between EDataAugmentation::SetMinContrastGain and EDataAugmentation::SetMaxContrastGain. A contrast transformation does not change the overall brightness of an image.

◆ GetMinGamma()

float eVision::EasyDeepLearning::EDataAugmentation::GetMinGamma ( ) const

Gets the minimum gamma for gamma correction.

Gamma correction transformation is performed by raising each normalized pixel value to the power of gamma with gamma taken between EDataAugmentation::SetMinGamma and EDataAugmentation::SetMaxGamma.

◆ GetMinSaturationGain()

float eVision::EasyDeepLearning::EDataAugmentation::GetMinSaturationGain ( ) const

Gets the minimum saturation gain.

◆ GetMinScale()

float eVision::EasyDeepLearning::EDataAugmentation::GetMinScale ( ) const

Gets the minimum scaling allowed.

◆ GetMinStainColor()

uint8_t eVision::EasyDeepLearning::EDataAugmentation::GetMinStainColor ( )

The minimum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).

Its value must be inferior to 255 and EDataAugmentation::SetMaxStainColor.

◆ GetSaltAndPepperNoiseMaximumDensity()

float eVision::EasyDeepLearning::EDataAugmentation::GetSaltAndPepperNoiseMaximumDensity ( ) const

The maximum density of the salt and pepper noise.

The salt and pepper noise sets the value of a number of randomly selected (between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and EDataAugmentation::SetSaltAndPepperNoiseMaximumDensity) pixels to its minimum or maximum value.

Its value must be between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and 1.

This noise is computed after all the other noises.

◆ GetSaltAndPepperNoiseMinimumDensity()

float eVision::EasyDeepLearning::EDataAugmentation::GetSaltAndPepperNoiseMinimumDensity ( ) const

The minimum density of the salt and pepper noise.

The salt and pepper noise sets the value of a number of randomly selected (between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and EDataAugmentation::SetSaltAndPepperNoiseMaximumDensity) pixels to its minimum or maximum value.

Its value must be between 0 and EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity.

This noise is computed after all the other noises.

◆ GetSpeckleNoiseMaximumStandardDeviation()

float eVision::EasyDeepLearning::EDataAugmentation::GetSpeckleNoiseMaximumStandardDeviation ( ) const

The speckle noise maximum standard deviation.

The speckle noise is a multiplicative noise sampled from a Gamma distribution of deviation between EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation and EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

Its value must be strictly higher than EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation.

This noise is computed before the salt and paper noise.

◆ GetSpeckleNoiseMinimumStandardDeviation()

float eVision::EasyDeepLearning::EDataAugmentation::GetSpeckleNoiseMinimumStandardDeviation ( ) const

The speckle noise minimum standard deviation.

The speckle noise is a multiplicative noise sampled from a Gamma distribution of deviation between EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation and EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

Its value must be strictly positive and lower than EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

This noise is computed before the salt and paper noise.

◆ GetStainBlur()

int eVision::EasyDeepLearning::EDataAugmentation::GetStainBlur ( ) const

Stain blur.

The stain blur is the half kernel size of a Gaussian filter that is applied on the stain to smooth its edges and therefore the transition between the original image and the stain.

◆ GetStainColorVariation()

uint32_t eVision::EasyDeepLearning::EDataAugmentation::GetStainColorVariation ( )

The variation of the color used to fill in the ellipse (We use an gaussian distribution of standard mean deviation stainColorVariation).

◆ GetStainDisruptionMaxAnchorPoints()

uint32_t eVision::EasyDeepLearning::EDataAugmentation::GetStainDisruptionMaxAnchorPoints ( )

The maximum number of anchor points to use while disrupting the ellipse.

◆ GetStainDisruptionMaxIntensity()

float eVision::EasyDeepLearning::EDataAugmentation::GetStainDisruptionMaxIntensity ( )

The maximum offset applied to the coordinates (x,y) of a pixel from the ellipse contour.

◆ GetStainEllipseMaxRadius()

uint32_t eVision::EasyDeepLearning::EDataAugmentation::GetStainEllipseMaxRadius ( )

The maximum radius of the ellipse used to produce the stain.

Its value must be superior to EDataAugmentation::SetStainEllipseMinRadius.

◆ GetStainEllipseMinRadius()

uint32_t eVision::EasyDeepLearning::EDataAugmentation::GetStainEllipseMinRadius ( )

The minimum radius of the ellipse used to produce the stain.

Its value must be inferior to EDataAugmentation::SetStainEllipseMaxRadius.

◆ GetStainInterpolationSiteFactor()

float eVision::EasyDeepLearning::EDataAugmentation::GetStainInterpolationSiteFactor ( )

The factor to compute the number of interpolation sites to use from the number of anchor points.

◆ GetStainProbability()

float eVision::EasyDeepLearning::EDataAugmentation::GetStainProbability ( )

Probability of applying the stain data augmentation (default value: 0, i.e. no stain).

◆ HasDataAugmentations()

bool eVision::EasyDeepLearning::EDataAugmentation::HasDataAugmentations ( ) const

Whether the EDataAugmentation object contains augmentations.

◆ Load() [1/2]

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

Loads an EDataAugmentation object. The given ESerializer must have been created for reading.

Parameters

path	The file path.

◆ Load() [2/2]

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

Loads an EDataAugmentation object. The given ESerializer must have been created for reading.

Parameters

serializer The serializer.

◆ operator=()

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

Copies the EDataAugmentation object other to this object.

Parameters

other The other EDataAugmentation that we will copy into this one.

◆ operator==()

bool eVision::EasyDeepLearning::EDataAugmentation::operator== ( const eVision::EasyDeepLearning::EDataAugmentation & other ) const

Equality operator.

Parameters

other Other object to compare to

◆ Save() [1/2]

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

Saves an EDataAugmentation object. The given ESerializer must have been created for writing.

Parameters

path	The file path.

◆ Save() [2/2]

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

Saves an EDataAugmentation object. The given ESerializer must have been created for writing.

Parameters

serializer The ESerializer object that is written to.

◆ SetEnableHorizontalFlip()

void eVision::EasyDeepLearning::EDataAugmentation::SetEnableHorizontalFlip ( bool enable )

Sets whether to use horizontally flipped versions of input images.

Parameters

enable Whether to enable horizontal flips.

◆ SetEnableVerticalFlip()

void eVision::EasyDeepLearning::EDataAugmentation::SetEnableVerticalFlip ( bool enable )

Sets whether to use vertically flipped versions of input images.

Parameters

enable Whether to enable vertical flips.

◆ SetGaussianNoiseMaximumStandardDeviation()

void eVision::EasyDeepLearning::EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation ( float gaussianMaximumDeviation )

The Gaussian noise maximum standard deviation.

The Gaussian noise is an additive noise sampled from a Gaussian distribution of deviation between EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.(also called the normal distribution).

Its value must be superior or equal to EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation.

Parameters

gaussianMaximumDeviation The maximum standard deviation of the gaussian distribution expressed in percentage of the image type max intensity

This noise is computed before the salt and paper noise.

◆ SetGaussianNoiseMinimumStandardDeviation()

void eVision::EasyDeepLearning::EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation ( float gaussianMinimumDeviation )

The Gaussian noise minimum standard deviation.

The Gaussian noise is an additive noise sampled from a Gaussian distribution of deviation between EDataAugmentation::SetGaussianNoiseMinimumStandardDeviation and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.(also called the normal distribution).

Its value must be between 0 and EDataAugmentation::SetGaussianNoiseMaximumStandardDeviation.

Parameters

gaussianMinimumDeviation The minimum standard deviation of the gaussian distribution expressed in percentage of the image type max intensity

This noise is computed before the salt and paper noise.

◆ SetMaxBrightnessOffset()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxBrightnessOffset ( float offset )

Sets the maximum absolute brightness offset. It must be between 0 and 1.

Parameters

offset The maximum absolute offset to the brightness

Brightness transformation is performed by adding a value taken between -EDataAugmentation::SetMaxBrightnessOffset and +EDataAugmentation::SetMaxBrightnessOffset to each pixel of the normalized image.

◆ SetMaxContrastGain()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxContrastGain ( float val )

Sets the maximum contrast gain. Its value must be strictly positive and over EDataAugmentation::SetMinContrastGain.

Parameters

val	The maximum contrast gain

Contrast transformation is performed by multiplying each mean-centered pixel value by a gain value taken between EDataAugmentation::SetMinContrastGain and EDataAugmentation::SetMaxContrastGain. A contrast transformation does not change the overall brightness of an image.

◆ SetMaxGamma()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxGamma ( float gamma )

Sets the maximum gamma for gamma correction. Its value must be higher than EDataAugmentation::SetMinGamma.

Parameters

gamma The maximum gamma

Gamma correction transformation is performed by raising each normalized pixel value to the power of gamma with gamma taken between EDataAugmentation::SetMinGamma and EDataAugmentation::SetMaxGamma.

◆ SetMaxHorizontalShear()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxHorizontalShear ( float hShear )

Sets the maximum absolute horizontal shear, represented as an angle from the vertical direction. Its value must be between 0 and 90 degrees.

Parameters

hShear The maximum horizontal shear allowed

◆ SetMaxHorizontalShift()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxHorizontalShift ( int32_t maxShift )

Sets the maximum horizontal shift allowed.

Parameters

maxShift Maximum horizontal shift

◆ SetMaxHueOffset()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxHueOffset ( float hueOffset )

Sets the maximum absolute hue offset. Its value must be between 0 and 180 degrees.

Parameters

hueOffset The maximum hue offset

The hue is represented as an angle between 0 and 360 degrees. The hue transformation is performed by rotating the hue of each pixel by a value between -EDataAugmentation::SetMaxHueOffset and +EDataAugmentation::SetMaxHueOffset. This transformation only works for color images.

◆ SetMaxRotationAngle()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxRotationAngle ( float maxAngle )

Set the maximum rotation angle allowed. Its value must be between 0 and 180 degrees.

The unit of this angle is in the current angle unit (see Easy::SetAngleUnit).

Parameters

maxAngle The maximum angle allowed.

◆ SetMaxSaturationGain()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxSaturationGain ( float saturationGain )

Sets the maximum saturation gain. Its value must be over or equal to EDataAugmentation::SetMinSaturationGain.

Parameters

saturationGain The maximum saturation gain

The saturation transformation is performed by multiplying the saturation of each pixel by a value between EDataAugmentation::SetMinSaturationGain and EDataAugmentation::SetMaxSaturationGain.

◆ SetMaxScale()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxScale ( float maxScale )

Sets the maximum scaling allowed. Its value must be strictly positive and over EDataAugmentation::SetMinScale.

Parameters

maxScale The maximum scaling allowed

◆ SetMaxStainColor()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxStainColor ( uint8_t maxStainColor )

The maximum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).

Its value must be bigger than EDataAugmentation::SetMinStainColor.

Parameters

maxStainColor The color

◆ SetMaxVerticalShear()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxVerticalShear ( float vShear )

Sets the maximum absolute vertical shear, represented as an angle from the horizontal direction. Its value must be between 0 and 90 degrees.

Parameters

vShear The maximum vertical shear allowed

◆ SetMaxVerticalShift()

void eVision::EasyDeepLearning::EDataAugmentation::SetMaxVerticalShift ( int32_t maxShift )

Sets the maximum vertical shift allowed.

◆ SetMinContrastGain()

void eVision::EasyDeepLearning::EDataAugmentation::SetMinContrastGain ( float val )

Sets the minimum contrast gain. Its value must be strictly positive and below EDataAugmentation::SetMaxContrastGain.

Parameters

val	The minimum contrast gain

Contrast transformation is performed by multiplying each mean-centered pixel value by a gain value taken between EDataAugmentation::SetMinContrastGain and EDataAugmentation::SetMaxContrastGain. A contrast transformation does not change the overall brightness of an image.

◆ SetMinGamma()

void eVision::EasyDeepLearning::EDataAugmentation::SetMinGamma ( float gamma )

Sets the minimum gamma for gamma correction. Its value must be strictly positive and below EDataAugmentation::SetMaxGamma.

Parameters

gamma The minimum gamma

Gamma correction transformation is performed by raising each normalized pixel value to the power of gamma with gamma taken between EDataAugmentation::SetMinGamma and EDataAugmentation::SetMaxGamma.

◆ SetMinSaturationGain()

void eVision::EasyDeepLearning::EDataAugmentation::SetMinSaturationGain ( float saturationGain )

Sets the minimum saturation gain. Its value must be strictly positive.

Parameters

saturationGain The minimum saturation gain

The saturation transformation is performed by multiplying the saturation of each pixel by a value between EDataAugmentation::SetMinSaturationGain and EDataAugmentation::SetMaxSaturationGain.

◆ SetMinScale()

void eVision::EasyDeepLearning::EDataAugmentation::SetMinScale ( float minScale )

Sets the minimum scaling allowed. Its value must be strictly positive and below EDataAugmentation::SetMaxScale.

Parameters

minScale The minimum scaling allowed

◆ SetMinStainColor()

void eVision::EasyDeepLearning::EDataAugmentation::SetMinStainColor ( uint8_t minStainColor )

The minimum color value from which to draw a color to fill in the stain (we use a gaussian distribution of with a mean between EDataAugmentation::SetMinStainColor and EDataAugmentation::SetMaxStainColor).

Its value must be inferior to 255 and EDataAugmentation::SetMaxStainColor.

Parameters

minStainColor The color

◆ SetSaltAndPepperNoiseMaximumDensity()

void eVision::EasyDeepLearning::EDataAugmentation::SetSaltAndPepperNoiseMaximumDensity ( float saltAndPepperNoiseMaximumDensity )

The maximum density of the salt and pepper noise.

The salt and pepper noise sets the value of a number of randomly selected (between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and EDataAugmentation::SetSaltAndPepperNoiseMaximumDensity) pixels to its minimum or maximum value.

Its value must be between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and 1.

Parameters

saltAndPepperNoiseMaximumDensity The maximum density of the salt and pepper noise expressed in percentage of the total number of pixels of the image

This noise is computed after all the other noises.

◆ SetSaltAndPepperNoiseMinimumDensity()

void eVision::EasyDeepLearning::EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity ( float saltAndPepperNoiseMinimumDensity )

The minimum density of the salt and pepper noise.

The salt and pepper noise sets the value of a number of randomly selected (between EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity and EDataAugmentation::SetSaltAndPepperNoiseMaximumDensity) pixels to its minimum or maximum value.

Its value must be between 0 and EDataAugmentation::SetSaltAndPepperNoiseMinimumDensity.

Parameters

saltAndPepperNoiseMinimumDensity The minimum density of the salt and pepper noise expressed in percentage of the total number of pixels of the image

This noise is computed after all the other noises.

◆ SetSpeckleNoiseMaximumStandardDeviation()

void eVision::EasyDeepLearning::EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation ( float speckleNoiseMaximumStandardDeviation )

The speckle noise maximum standard deviation.

The speckle noise is a multiplicative noise sampled from a Gamma distribution of deviation between EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation and EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

Its value must be strictly higher than EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation.

Parameters

speckleNoiseMaximumStandardDeviation The standard maximum deviation of the gamma distribution

This noise is computed before the salt and paper noise.

◆ SetSpeckleNoiseMinimumStandardDeviation()

void eVision::EasyDeepLearning::EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation ( float speckleNoiseMinimumStandardDeviation )

The speckle noise minimum standard deviation.

The speckle noise is a multiplicative noise sampled from a Gamma distribution of deviation between EDataAugmentation::SetSpeckleNoiseMinimumStandardDeviation and EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

Its value must be strictly positive and lower than EDataAugmentation::SetSpeckleNoiseMaximumStandardDeviation.

Parameters

speckleNoiseMinimumStandardDeviation The standard minimum deviation of the gamma distribution

This noise is computed before the salt and paper noise.

◆ SetStainBlur()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainBlur ( int blur )

Stain blur.

The stain blur is the half kernel size of a Gaussian filter that is applied on the stain to smooth its edges and therefore the transition between the original image and the stain.

Parameters

blur	Blur to apply to the stain

◆ SetStainColorVariation()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainColorVariation ( uint32_t stainColorVariation )

The variation of the color used to fill in the ellipse (We use an gaussian distribution of standard mean deviation stainColorVariation).

Parameters

stainColorVariation The variation of the color

◆ SetStainDisruptionMaxAnchorPoints()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainDisruptionMaxAnchorPoints ( uint32_t stainDisruptioMaxAnchorPoints )

The maximum number of anchor points to use while disrupting the ellipse.

◆ SetStainDisruptionMaxIntensity()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainDisruptionMaxIntensity ( float stainDisruptionMaxIntensity )

The maximum offset applied to the coordinates (x,y) of a pixel from the ellipse contour.

Parameters

stainDisruptionMaxIntensity The maximum offset

◆ SetStainEllipseMaxRadius()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainEllipseMaxRadius ( uint32_t stainEllipseMaxRadius )

The maximum radius of the ellipse used to produce the stain.

Its value must be superior to EDataAugmentation::SetStainEllipseMinRadius.

Parameters

stainEllipseMaxRadius The maximum radius

◆ SetStainEllipseMinRadius()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainEllipseMinRadius ( uint32_t stainEllipseMinRadius )

The minimum radius of the ellipse used to produce the stain.

Its value must be inferior to EDataAugmentation::SetStainEllipseMaxRadius.

Parameters

stainEllipseMinRadius The minimum radius

◆ SetStainInterpolationSiteFactor()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainInterpolationSiteFactor ( float interpolationSiteFactor )

The factor to compute the number of interpolation sites to use from the number of anchor points.

Parameters

interpolationSiteFactor Interpolation site factor.

◆ SetStainProbability()

void eVision::EasyDeepLearning::EDataAugmentation::SetStainProbability ( float stainProbability )

Probability of applying the stain data augmentation (default value: 0, i.e. no stain).

Parameters

stainProbability Probability of applying stain data augmentation (between 0 and 1).

Member Data Documentation

◆ wrappedClass_

CppWrapperInternal::WrappedClass eVision::EasyDeepLearning::EDataAugmentation::wrappedClass_

Public Member Functions

Public Attributes

Detailed Description

Constructor & Destructor Documentation

◆ EDataAugmentation() [1/3]

◆ EDataAugmentation() [2/3]

◆ ~EDataAugmentation()

◆ EDataAugmentation() [3/3]

Member Function Documentation

◆ CopyTo()

◆ Generate() [1/8]

◆ Generate() [2/8]

◆ Generate() [3/8]

◆ Generate() [4/8]

◆ Generate() [5/8]

◆ Generate() [6/8]

◆ Generate() [7/8]

◆ Generate() [8/8]

◆ GetEnableHorizontalFlip()

◆ GetEnableVerticalFlip()

◆ GetGaussianNoiseMaximumStandardDeviation()

◆ GetGaussianNoiseMinimumStandardDeviation()

◆ GetMaxBrightnessOffset()

◆ GetMaxContrastGain()

◆ GetMaxGamma()

◆ GetMaxHorizontalShear()

◆ GetMaxHorizontalShift()

◆ GetMaxHueOffset()

◆ GetMaxRotationAngle()

◆ GetMaxSaturationGain()

◆ GetMaxScale()

◆ GetMaxStainColor()

◆ GetMaxVerticalShear()

◆ GetMaxVerticalShift()

◆ GetMinContrastGain()

◆ GetMinGamma()

◆ GetMinSaturationGain()

◆ GetMinScale()

◆ GetMinStainColor()

◆ GetSaltAndPepperNoiseMaximumDensity()

◆ GetSaltAndPepperNoiseMinimumDensity()

◆ GetSpeckleNoiseMaximumStandardDeviation()

◆ GetSpeckleNoiseMinimumStandardDeviation()

◆ GetStainBlur()

◆ GetStainColorVariation()

◆ GetStainDisruptionMaxAnchorPoints()

◆ GetStainDisruptionMaxIntensity()

◆ GetStainEllipseMaxRadius()

◆ GetStainEllipseMinRadius()

◆ GetStainInterpolationSiteFactor()

◆ GetStainProbability()

◆ HasDataAugmentations()

◆ Load() [1/2]

◆ Load() [2/2]

◆ operator=()

◆ operator==()

◆ Save() [1/2]

◆ Save() [2/2]

◆ SetEnableHorizontalFlip()

◆ SetEnableVerticalFlip()

◆ SetGaussianNoiseMaximumStandardDeviation()

◆ SetGaussianNoiseMinimumStandardDeviation()

◆ SetMaxBrightnessOffset()

◆ SetMaxContrastGain()

◆ SetMaxGamma()

◆ SetMaxHorizontalShear()

◆ SetMaxHorizontalShift()

◆ SetMaxHueOffset()

◆ SetMaxRotationAngle()

◆ SetMaxSaturationGain()

◆ SetMaxScale()

◆ SetMaxStainColor()

◆ SetMaxVerticalShear()

◆ SetMaxVerticalShift()

◆ SetMinContrastGain()

◆ SetMinGamma()

◆ SetMinSaturationGain()

◆ SetMinScale()

◆ SetMinStainColor()

◆ SetSaltAndPepperNoiseMaximumDensity()