test_repeatability.cc File Reference

---

Detailed Description

Main file for the test_repeatability executable.

Usage

test_repeatability [--var VAL] [--exec FILE]

Description

This program loads a dataset and then computes repeatability of the specified detector on the dataset. This program accepts standard GVars3 commandline arguments and loads learn_detector.cfg as a the default configuration:

////////////////////////////////////////////////////////////////////////////////
//
// Parameters for various corner detectors
//

//Parameters for the Harris and Shi-Tomasi detector
harris.blur=2.5        //Standard deviation of blur
harris.sigmas=2.0      //Blur to this many standard deviations

shitomasi.blur=2.5
shitomasi.sigmas=2.0

//Parameters for DoG detector
dog.sigma=1.0                //Initial blur
dog.divisions_per_octave=3  
dog.octaves=4                

//Paramaters for Harris-Laplace detector
harrislaplace.harris.blur=1
harrislaplace.harris.sigmas=2
harrislaplace.dog.sigma=1
harrislaplace.dog.divisions_per_octave=7
harrislaplace.dog.octaves=4

//SUSAN detector
susan.dt=4

///Parameters for FAST-ER tree detector.
offsets.min_radius=2.0  //This must be the same as the value used in training
offsets.max_radius=4.2  //This must be the same as the value used in training

debug.verify_detections=0 //Debugging code. Leave to 0.
debug.verify_scores=0

faster2=best_faster.tree  //Detector file to load

////////////////////////////////////////////////////////////////////////////////
//
// Parameters for the experiment
// 

// Number of corners per frame to use
cpf=0 10 20 30 40 50 60 70 80 90 100 150 200 250 300 350 400 450 500 550 600 650 700 750 800 850 900 950 1000 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000 2200

ncpf=500      //Number of corners per frame to use for the noise test
nmax=50       //Noise standard deviation to run to in the noise test
r=5           //Radius used to determine if the point is repeated
test="normal" //Type of test to run. Options are normal or noise

The available detectors are selected using the detector variable. Options are given in get_detector.

Definition in file test_repeatability.cc.

#include <iostream>
#include <sstream>
#include <cfloat>
#include <map>
#include <utility>
#include <cvd/image_io.h>
#include <cvd/random.h>
#include <cvd/image_interpolate.h>
#include <tag/printf.h>
#include <tag/stdpp.h>
#include "gvars_vector.h"
#include "load_data.h"
#include "detectors.h"
#include "utility.h"

Go to the source code of this file.

Functions
double	compute_repeatability_exact (const vector< vector< Image< array< float, 2 > > > > &warps, const vector< vector< ImageRef > > &corners, double r)
void	compute_repeatability_all (const vector< Image< byte > > &images, const vector< vector< Image< array< float, 2 > > > > &warps, const DetectN &detector, const vector< int > &cpf, double fuzz)
void	compute_repeatability_noise (const vector< Image< byte > > &images, const vector< vector< Image< array< float, 2 > > > > &warps, const DetectN &detector, int cpf, float n, double fuzz)
void	mmain (int argc, char **argv)
int	main (int argc, char **argv)

---

Function Documentation

int main	(	int	argc,
		char **	argv
	)

Driving function which catches exceptions.

Parameters:

	argc	Number of command line arguments
	argv	Commandline argument list

Definition at line 231 of file test_repeatability.cc.

References mmain().

{
    try
    {
        mmain(argc, argv);
    }
    catch(Exceptions::All e)
    {
        cerr << "Error: " << e.what << endl;
    }   
}

---