BootStrapProcedure.h File Reference

methods to evaluate classifier on images and detect false positives More...

#include "NonMaximaSuppression.h"
#include "Candidate.h"
#include "IO/datasetin.h"
#include "Utility/ResponseUtils.h"

Go to the source code of this file.

Classes
struct	BootStrapParams
	Parameters for BootStrapProcedure. More...

Functions
template<class SourceDatasetType , class ClassifierType >
bool	BootStrapProcedure (SourceDatasetType &training_set, datasetin &in, const ClassifierType &cl, const BootStrapParams &params, int num_thread)

Variables
int	frameidx
	a counter, used for debug More...

Detailed Description

methods to evaluate classifier on images and detect false positives

Function Documentation

template<class SourceDatasetType , class ClassifierType >

bool BootStrapProcedure	(	SourceDatasetType &	training_set,
		datasetin &	in,
		const ClassifierType &	cl,
		const BootStrapParams &	params,
		int	num_thread
	)

select best negative greater then thMin in list of image in filename TODO nNegImage

Parameters

training_set
in	un dataset
cl	a classifier
thMin	minimum threshold to accept negative
nNegImage	number of negatives per frame

Variable Documentation

int frameidx

a counter, used for debug

Todo:

remove it