Base abstract class for node potentials training. More...

#include <TrainNode.h>

Inheritance diagram for DirectGraphicalModels::CTrainNode:

[legend]

Collaboration diagram for DirectGraphicalModels::CTrainNode:

Public Member Functions
	CTrainNode (byte nStates, word nFeatures)
	Constructor. More...

virtual	~CTrainNode (void)=default

void	addFeatureVecs (const Mat &featureVectors, const Mat &gt)
	Adds a block of new feature vectors. More...

void	addFeatureVecs (const vec_mat_t &featureVectors, const Mat &gt)
	Adds a block of new feature vectors. More...

virtual void	addFeatureVec (const Mat &featureVector, byte gt)=0
	Adds new feature vector. More...

virtual void	train (bool doClean=false)
	Random model training. More...

Mat	getNodePotentials (const Mat &featureVectors, const Mat &weights=Mat(), float Z=0.0f) const
	Returns a block of node potentials, based on the block of feature vector. More...

Mat	getNodePotentials (const vec_mat_t &featureVectors, const Mat &weights=Mat(), float Z=0.0f) const
	Returns a block of node potentials, based on the block of feature vector. More...

Mat	getNodePotentials (const Mat &featureVector, float weight, float Z=0.0f) const
	Returns the node potential, based on the feature vector. More...

Public Member Functions inherited from DirectGraphicalModels::ITrain
	ITrain (byte nStates, word nFeatures)
	Constructor. More...

virtual	~ITrain (void)=default

word	getNumFeatures (void) const
	Returns number of features. More...

Public Member Functions inherited from DirectGraphicalModels::CBaseRandomModel
	CBaseRandomModel (byte nStates)
	Constructor. More...

virtual	~CBaseRandomModel (void)

virtual void	reset (void)=0
	Resets class variables. More...

virtual void	save (const std::string &path, const std::string &name=std::string(), short idx=-1) const
	Saves the training data. More...

virtual void	load (const std::string &path, const std::string &name=std::string(), short idx=-1)
	Loads the training data. More...

byte	getNumStates (void) const
	Returns number of states (classes) More...

Static Public Member Functions
static std::shared_ptr< CTrainNode >	create (byte nodeRandomModel, byte nStates, word nFeatures)
	Factory method returning node trainer object. More...

Protected Member Functions
virtual void	calculateNodePotentials (const Mat &featureVector, Mat &potential, Mat &mask) const =0
	Calculates the node potential, based on the feature vector. More...

Protected Member Functions inherited from DirectGraphicalModels::CBaseRandomModel
virtual void	saveFile (FILE *pFile) const =0
	Saves the random model into the file. More...

virtual void	loadFile (FILE *pFile)=0
	Loads the random model from the file. More...

std::string	generateFileName (const std::string &path, const std::string &name, short idx) const
	Generates name of the data file for storing random model parameters. More...

Private Attributes
Mat	m_mask

Additional Inherited Members
Protected Attributes inherited from DirectGraphicalModels::CBaseRandomModel
byte	m_nStates
	The number of states (classes) More...

Detailed Description

Base abstract class for node potentials training.

The common usage for node potentials training is as follow:

CTrainNode *t = new CTrainNodeXXX(nStates, nFeatures);
// "adding data" phase 
for (int i = 0; i < NUMBER_OF_TRAINING_DATA; i++) t->addFeatureVec(train_featurVector[i], groundtruth_class[i]);
// training
t->train();
// "getting data" phase
for (int i = 0; i < NUMBER_OF_TEST_DATA; i++) predicted_class[i] = t->getNodePotentials(test_featurVector[i]);
delete t;

See Demo Train for more details

Author: Sergey G. Kosov, serge.nosp@m.y.ko.nosp@m.sov@p.nosp@m.roje.nosp@m.ct-10.nosp@m..de

Definition at line 47 of file TrainNode.h.

Constructor & Destructor Documentation

◆ CTrainNode()

DirectGraphicalModels::CTrainNode::CTrainNode	(	byte	nStates,
		word	nFeatures
	)

inline

Constructor.

Parameters

nStates	Number of states (classes)
nFeatures	Number of features

Definition at line 55 of file TrainNode.h.

◆ ~CTrainNode()

virtual DirectGraphicalModels::CTrainNode::~CTrainNode ( void )

virtualdefault

Member Function Documentation

◆ addFeatureVec()

virtual void DirectGraphicalModels::CTrainNode::addFeatureVec	(	const Mat &	featureVector,
		byte	gt
	)

pure virtual

Adds new feature vector.

Used to add a featureVector, corresponding to the ground-truth state (class) gt for training

Parameters

featureVector	Multi-dimensinal point: Mat(size: nFeatures x 1; type: CV_8UC1)
gt	Corresponding ground-truth state (class)

Implemented in DirectGraphicalModels::CTrainNodeMsRF, DirectGraphicalModels::CTrainNodeCvRF, DirectGraphicalModels::CTrainNodeCvANN, DirectGraphicalModels::CTrainNodeCvGMM, DirectGraphicalModels::CTrainNodeCvSVM, DirectGraphicalModels::CTrainNodeKNN, DirectGraphicalModels::CTrainNodeGMM, DirectGraphicalModels::CTrainNodeCvKNN, and DirectGraphicalModels::CTrainNodeBayes.

Here is the caller graph for this function:

◆ addFeatureVecs() [1/2]

void DirectGraphicalModels::CTrainNode::addFeatureVecs	(	const Mat &	featureVectors,
		const Mat &	gt
	)

Adds a block of new feature vectors.

Used to add multiple featureVectors, corresponding to the ground-truth states (classes) gt for training

Parameters

featureVectors	Multi-channel matrix, each element of which is a multi-dimensinal point: Mat(type: CV_8UC(nFeatures))
gt	Matrix, each element of which is a ground-truth state (class)

Definition at line 42 of file TrainNode.cpp.

Here is the call graph for this function:

◆ addFeatureVecs() [2/2]

void DirectGraphicalModels::CTrainNode::addFeatureVecs	(	const vec_mat_t &	featureVectors,
		const Mat &	gt
	)

Adds a block of new feature vectors.

Used to add multiple featureVectors, corresponding to the ground-truth states (classes) gt for training

Parameters

featureVectors	Vector of size nFeatures, each element of which is a single feature - image: Mat(type: CV_8UC1)
gt	Matrix, each element of which is a ground-truth state (class)

Definition at line 48 of file TrainNode.cpp.

Here is the call graph for this function:

◆ calculateNodePotentials()

virtual void DirectGraphicalModels::CTrainNode::calculateNodePotentials	(	const Mat &	featureVector,
		Mat &	potential,
		Mat &	mask
	)		const

protectedpure virtual

Calculates the node potential, based on the feature vector.

This function calculates the potentials of the node, described with the sample featureVector, being in each state (belonging to each class). These potentials are united in the node potential vector:

\[nodePot[nStates] = f(\textbf{f}[nFeatures]).\]

Functions \( f \) must be implemented in derived classes.

Parameters

[in]	featureVector	Multi-dimensinal point \(\textbf{f}\): Mat(size: nFeatures x 1; type: CV_{XX}C1)
[in,out]	potential	Node potentials: Mat(size: nStates x 1; type: CV_32FC1). This parameter should be preinitialized and set to value 0.
[in,out]	mask	Relevant Node potentials: Mat(size: nStates x 1; type: CV_8UC1). This parameter should be preinitialized and set to value 1 (all potentials are relevant).

Implemented in DirectGraphicalModels::CTrainNodeMsRF, DirectGraphicalModels::CTrainNodeCvRF, DirectGraphicalModels::CTrainNodeGMM, DirectGraphicalModels::CTrainNodeCvANN, DirectGraphicalModels::CTrainNodeCvGMM, DirectGraphicalModels::CTrainNodeCvSVM, DirectGraphicalModels::CTrainNodeBayes, DirectGraphicalModels::CTrainNodeKNN, and DirectGraphicalModels::CTrainNodeCvKNN.

Here is the caller graph for this function:

◆ create()

std::shared_ptr< CTrainNode > DirectGraphicalModels::CTrainNode::create	(	byte	nodeRandomModel,
		byte	nStates,
		word	nFeatures
	)

static

Factory method returning node trainer object.

Note: The resulting node trainer object is created with default parameters

Parameters

nodeRandomModel	Type of desired random model (Ref. NodeRandomModel)
nStates	Number of states (classes)
nFeatures	Number of features

Returns: Tne pointer to the concrete implementation of the node trainer class

Definition at line 20 of file TrainNode.cpp.

◆ getNodePotentials() [1/3]

Mat DirectGraphicalModels::CTrainNode::getNodePotentials	(	const Mat &	featureVectors,
		const Mat &	weights = `Mat()`,
		float	Z = `0.0f`
	)		const

Returns a block of node potentials, based on the block of feature vector.

Parameters

featureVectors	Multi-channel matrix, each element of which is a multi-dimensinal point: Mat(type: CV_8UC(nFeatures))
weights	The block of weighting parameters Mat(type: CV_32FC1). If empty, values 1 are used.
Z	The value of partition function. In order to convert potential to the probability, it is multiplied by \(1/Z\). If \(Z\leq0\), the resulting node potentials are normalized to 100, independently for each potential.

Definition at line 54 of file TrainNode.cpp.

Here is the call graph for this function:

Here is the caller graph for this function:

◆ getNodePotentials() [2/3]

Mat DirectGraphicalModels::CTrainNode::getNodePotentials	(	const vec_mat_t &	featureVectors,
		const Mat &	weights = `Mat()`,
		float	Z = `0.0f`
	)		const

Returns a block of node potentials, based on the block of feature vector.

Parameters

featureVectors	Vector of size nFeatures, each element of which is a single feature - image: Mat(type: CV_8UC1)
weights	The block of weighting parameter Mat(type: CV_32FC1)
Z	The value of partition function. In order to convert potential to the probability, it is multiplied by \(1/Z\). If \(Z\leq0\), the resulting node potentials are normalized to 100, independently for each potential.

Definition at line 91 of file TrainNode.cpp.

Here is the call graph for this function:

◆ getNodePotentials() [3/3]

Mat DirectGraphicalModels::CTrainNode::getNodePotentials	(	const Mat &	featureVector,
		float	weight,
		float	Z = `0.0f`
	)		const

Returns the node potential, based on the feature vector.

This function calls calculateNodePotentials() function, which should be implemented in derived classes. After that, the resulting node potential is powered by parameter weight.

Parameters

featureVector	Multi-dimensinal point \(\textbf{f}\): Mat(size: nFeatures x 1; type: CV_{XX}C1)
weight	The weighting parameter (default value is 1)
Z	The value of partition function. In order to convert potential to the probability, it is multiplied by \(1/Z\). If \(Z\leq0\), the resulting node potentials are normalized to 100, independently for every function call.

Returns: Normalized node potentials on success: Mat(size: nStates x 1; type: CV_32FC1);

Definition at line 129 of file TrainNode.cpp.

Here is the call graph for this function:

◆ train()

virtual void DirectGraphicalModels::CTrainNode::train ( bool doClean = false )

inlinevirtual

Random model training.

Auxilary function for training - some derived classes may use this function inbetween training and classification phases

Note: This function must be called inbetween the training and classification phases

Parameters

doClean Flag indicating if the memory, keeping the trining data should be released after training

Implements DirectGraphicalModels::ITrain.

Reimplemented in DirectGraphicalModels::CTrainNodeMsRF, DirectGraphicalModels::CTrainNodeCvRF, DirectGraphicalModels::CTrainNodeCvANN, DirectGraphicalModels::CTrainNodeCvGMM, DirectGraphicalModels::CTrainNodeCvSVM, DirectGraphicalModels::CTrainNodeKNN, DirectGraphicalModels::CTrainNodeCvKNN, DirectGraphicalModels::CTrainNodeGMM, and DirectGraphicalModels::CTrainNodeBayes.

Definition at line 92 of file TrainNode.h.

Here is the caller graph for this function:

Member Data Documentation

◆ m_mask

Mat DirectGraphicalModels::CTrainNode::m_mask

private

Definition at line 139 of file TrainNode.h.

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

D:/Projects/DGM/modules/DGM/TrainNode.h
D:/Projects/DGM/modules/DGM/TrainNode.cpp

Public Member Functions

Static Public Member Functions

Protected Member Functions

Private Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ CTrainNode()

◆ ~CTrainNode()

Member Function Documentation

◆ addFeatureVec()

◆ addFeatureVecs() [1/2]

◆ addFeatureVecs() [2/2]

◆ calculateNodePotentials()

◆ create()

◆ getNodePotentials() [1/3]

◆ getNodePotentials() [2/3]

◆ getNodePotentials() [3/3]

◆ train()

Member Data Documentation

◆ m_mask