SVMClassifier Class Reference

SVM Classifier Class. More...

#include <Learn/SVMClassifier.H>

Collaboration diagram for SVMClassifier:

[legend]

List of all members.

Public Member Functions
	SVMClassifier (float gamma=0.00078125, int C=32)
	Constructor.
	~SVMClassifier ()
	Destructor.
void	train (std::string outputFileName, int id, std::vector< float > &feature)
	Train.
void	train (std::string outputFileName, int id, float *&feature, unsigned int fdim)
void	train (std::string outputFilename, int id, float **&feature, unsigned int fdim1, unsigned int fdim2)
void	train (Image< double > trainingData, std::vector< double > dataClasses)
	Train the SVM using the data in trainingData, with the ground truth data in dataClasses.
void	train (std::vector< std::vector< float > > trainingDta, std::vector< float > dataClasses)
std::map< int, double >	predictPDF (const svm_node *dataPointNodes)
	Predict the class of a data point.
std::map< int, double >	predictPDF (std::vector< float > &feature)
double	predict (Image< double > dataPoint, double *probability=NULL)
	Predict.
double	predict (std::vector< float > &feature, double *probability=NULL)
double	predict (float *&feature, unsigned int fdim, double *probability=NULL)
double	predict (float **&feature, unsigned int fdim1, unsigned int fdim2, double *probability=NULL)
void	readModel (std::string modelFileName)
void	writeModel (std::string modelFileName)
	Read model file into memory.
void	readRange (std::string rangeFileName)
	Write model file into memory.
float	rescaleValue (float value, unsigned int index)
	Read a range file to scale the feature vector, if needed.
Protected Member Functions
int	_getBestLabel (std::map< int, double > pdf, double *probability)
	If range is enabled, rescale the value to.
double	_predict (struct svm_node *node, unsigned int fdim, double *probability)
void	_train (svm_problem &trainingProblem)
Protected Attributes
struct svm_parameter	itsSVMParams
struct svm_model *	itsSVMModel
bool	itsSVMRangeEnabled
std::vector< double >	itsSVMFeatureRangeMax
std::vector< double >	itsSVMFeatureRangeMin
double	itsSVMFeatureRangeUpper
double	itsSVMFeatureRangeLower

---

Detailed Description

SVM Classifier Class.

Definition at line 53 of file SVMClassifier.H.

---

Constructor & Destructor Documentation

SVMClassifier::SVMClassifier	(	float	gamma = 0.00078125,
		int	C = 32
	)

Constructor.

Definition at line 52 of file SVMClassifier.C.

SVMClassifier::~SVMClassifier

(

)

Destructor.

Definition at line 75 of file SVMClassifier.C.

---

Member Function Documentation

int SVMClassifier::_getBestLabel	(	std::map< int, double >	pdf,
		double *	probability
	)			[protected]

If range is enabled, rescale the value to.

Definition at line 297 of file SVMClassifier.C.

Referenced by predict().

double SVMClassifier::predict	(	Image< double >	dataPoint,
		double *	probability = NULL
	)

Predict.

Definition at line 267 of file SVMClassifier.C.

References _getBestLabel(), ASSERT, Image< T >::getHeight(), Image< T >::getVal(), Image< T >::getWidth(), predictPDF(), and rescaleValue().

std::map< int, double > SVMClassifier::predictPDF

(

const svm_node *

dataPointNodes

)

Predict the class of a data point.

Definition at line 248 of file SVMClassifier.C.

Referenced by predict().

void SVMClassifier::readRange

(

std::string

rangeFileName

)

Write model file into memory.

Definition at line 87 of file SVMClassifier.C.

Referenced by SVMClassifierModule::start2().

float SVMClassifier::rescaleValue	(	float	value,
		unsigned int	index
	)

Read a range file to scale the feature vector, if needed.

Definition at line 150 of file SVMClassifier.C.

Referenced by predict().

void SVMClassifier::train	(	Image< double >	trainingData,
		std::vector< double >	dataClasses
	)

Train the SVM using the data in trainingData, with the ground truth data in dataClasses.

Definition at line 373 of file SVMClassifier.C.

References ASSERT, Image< T >::getHeight(), Image< T >::getVal(), and Image< T >::getWidth().

void SVMClassifier::train	(	std::string	outputFileName,
		int	id,
		std::vector< float > &	feature
	)

Train.

Definition at line 316 of file SVMClassifier.C.

void SVMClassifier::writeModel

(

std::string

modelFileName

)

Read model file into memory.

Definition at line 145 of file SVMClassifier.C.

---

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

• src/Learn/SVMClassifier.H
• src/Learn/SVMClassifier.C