ARLK::SVMClassifier Class Reference

#include <SVMClassifier.h>

Inheritance diagram for ARLK::SVMClassifier:

Public Member Functions
	SVMClassifier ()
	~SVMClassifier ()
void	SetSVMParameters (double C, double gamma, double weight)
virtual bool	FeatureExtraction (SkeData *inputData, const double *dataLabels=NULL, int labelNum=0)
virtual bool	train ()
	Create the svm model and save it as svm_model.
virtual double *	predict (SkeData *inputData, int *pLength)
virtual void	CleanData ()
	Clean all the data except parameters and set data number to 0.
Public Member Functions inherited from ARLK::BaseClassifier
	BaseClassifier ()
	~BaseClassifier ()

Additional Inherited Members
Static Public Member Functions inherited from ARLK::BaseClassifier
static int **	genConfusionMatrix (double *predictedLabels, double *trueLabels, int num, int labelTypeNum)
Static Protected Member Functions inherited from ARLK::BaseClassifier
static bool	CalFeatureVector (SkeData *inputData, int frameIdx, int vecIdx, FEATURE_VEC *pSkdPointOut, double wt)
Protected Attributes inherited from ARLK::BaseClassifier
int	dataNum
	Number of the data.
double *	labels
double	actWeight

Constructor & Destructor Documentation

ARLK::SVMClassifier::SVMClassifier

(

)

ARLK::SVMClassifier::~SVMClassifier

(

)

Member Function Documentation

void ARLK::SVMClassifier::CleanData ( )

virtual

Clean all the data except parameters and set data number to 0.

Reimplemented from ARLK::BaseClassifier.

bool ARLK::SVMClassifier::FeatureExtraction ( SkeData *  inputData, const double *  dataLabels = NULL, int  labelNum = 0  )

virtual

Extract features from SkeData

Parameters

[in]	inputData	SkeData containing all training data information
[in]	dataLabels	An array containing the data labels for each training data
[in]	labelNum	Number of labels/training data

Returns

True if succeeds

Reimplemented from ARLK::BaseClassifier.

double * ARLK::SVMClassifier::predict ( SkeData *  inputData, int *  pLength  )

virtual

Classify the data in inputData, store the results in the buffer pointed by labelBuf

Parameters

[in]	inputData	SkeData containing all test data information
[out]	pLength	Pointer to the integer indicating the length of the buffer of the predicted labels

Returns

A pointer to the buffer containing the predicted double labels, should be deleted by the user

Reimplemented from ARLK::BaseClassifier.

void ARLK::SVMClassifier::SetSVMParameters	(	double	C,
		double	gamma,
		double	weight
	)

Set SVM training parameters By default, C = 10, gamma = 0.001, weight = 3

Parameters

[in]	C	SVM training parameter C in svm_parameter
[in]	gamma	SVM training parameter gamma in svm_parameter
[in]	weight	Factor of the action features

bool ARLK::SVMClassifier::train ( )

virtual

Create the svm model and save it as svm_model.

Reimplemented from ARLK::BaseClassifier.

---

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

• include/ARLK/SVMClassifier.h
• src/ARLK/SVMClassifier.cpp