lti::classifier2DVisualizer Class Reference

This functor creates beautiful pictures. More...

#include <ltiClassifier2DVisualizer.h>

Inheritance diagram for lti::classifier2DVisualizer:

[legend]

Collaboration diagram for lti::classifier2DVisualizer:

[legend]

List of all members.

Classes
class	parameters
	the parameters for the class classifier2DVisualizer More...
Public Member Functions
	classifier2DVisualizer ()
	classifier2DVisualizer (const classifier2DVisualizer &other)
virtual	~classifier2DVisualizer ()
virtual const char *	getTypeName () const
bool	apply (const supervisedInstanceClassifier &cls, image &result) const
bool	apply (const supervisedInstanceClassifier &cls, const lti::dmatrix &highlights, image &result) const
bool	apply (const supervisedInstanceClassifier &cls, const lti::dmatrix &highlights, image &result, const ivector &labels) const
bool	apply (const supervisedInstanceClassifier &cls, const lti::dmatrix &highlights, image &result, const ivector &labels, const genericVector< bool > &htype) const
classifier2DVisualizer &	copy (const classifier2DVisualizer &other)
classifier2DVisualizer &	operator= (const classifier2DVisualizer &other)
virtual functor *	clone () const
const parameters &	getParameters () const

---

Detailed Description

This functor creates beautiful pictures.

They additionally have the property that they show the classification properties of a supervisedInstanceClassifier for a 2D feature space. Its application is rather simple: You have to create a classifier for a 2D space, then you give it to one of the apply methods, together with the target image.

Example:

    // number of training samples
 const int N=150;
 // number of classes
 const int K=3;
 const int width=400;
 const int height=200;

 // fill data and label matrix
 lti::dmatrix data(N,2);
 lti::ivector labels(N);
 int j=0;

 std::cerr << "Generating data...";
 for (int k=0; k<K; k++) {
   // generate gaussian
   double sigma=double(k)/(K)*20.0+10;
   double mx=double(k)/K*(width-sigma)+sigma;
   double my=double(k)/K*(height-sigma)+sigma;
   lti::gaussianDistribution px(mx,sigma);
   lti::gaussianDistribution py(my,sigma);
   for (int i=0; i<N/K; i++) {
     data[j][0]=px.draw();
     data[j][1]=py.draw();
     labels[j]=k;
     j++;
   }
 }

 std::cerr << "\nTraining MLP...";
 // train MLP
 lti::MLP s;
 lti::streamProgressInfo inf(std::cerr,"MLP");
 s.setProgressObject(inf);
 lti::MLP::parameters p=s.getParameters();

 s.setParameters(p);
 if (!s.train(data,labels)) {
   std::cerr << "Error: " << s.getStatusString() << "\n";
   //return;
 }

 std::cerr << "\nVisualizing...";
 // NOW COMES THE INTERESTING PART:
 // we need an image
 lti::image panel;
 lti::classifier2DVisualizer visual;
 lti::classifier2DVisualizer::parameters vp;
 // set the parameters
 vp.showBoundaries=true;
 vp.imgSize=point(width,height);
 vp.upperRight=dpoint(width,height);
 visual.setParameters(vp);
 // create the image
 if (!visual.apply(s,data,panel,labels)) {
   std::cerr << "Error: " << s.getStatusString() << "\n";
   return;
 }
 // enjoy!

The visualization result of this example is shown in

---

Constructor & Destructor Documentation

lti::classifier2DVisualizer::classifier2DVisualizer

(

)

default constructor

lti::classifier2DVisualizer::classifier2DVisualizer

(

const classifier2DVisualizer &

other

)

copy constructor

Parameters:

other

the object to be copied

virtual lti::classifier2DVisualizer::~classifier2DVisualizer

(

)

[virtual]

destructor

---

Member Function Documentation

bool lti::classifier2DVisualizer::apply	(	const supervisedInstanceClassifier &	cls,
		const lti::dmatrix &	highlights,
		image &	result,
		const ivector &	labels,
		const genericVector< bool > &	htype
	)			const

Applies the given classifier to the area defined in the parameters object.

The image will contain the visualization result. It also highlights the vectors given in highlights. The color of each highlight is determined by the corresponding value in labels. You can also distinguish between two classes of labels. The label class is determined by the vector htype.

Parameters:

	cls	supervisedInstanceClassifier with the classifier to be visualized. The classifier must be able to classify a 2D feature set.
	highlights	matrix with each row containing a vector that is to be highlighted.
	result	will receive the visualization result.
	labels	is a vector with the class labels for each row of highlights.
	htype	is a vector with a flag denoting the highlight type.

Returns:

true if apply successful or false otherwise.

bool lti::classifier2DVisualizer::apply	(	const supervisedInstanceClassifier &	cls,
		const lti::dmatrix &	highlights,
		image &	result,
		const ivector &	labels
	)			const

Applies the given classifier to the area defined in the parameters object.

The image will contain the visualization result. It also highlights the vectors given in highlights. The color of each highlight is determined by the corresponding value in labels.

Parameters:

	cls	supervisedInstanceClassifier with the classifier to be visualized. The classifier must be able to classify a 2D feature set.
	highlights	matrix with each row containing a vector that is to be highlighted.
	result	will receive the visualization result.
	labels	is a vector with the class labels for each row of highlights.

Returns:

true if apply successful or false otherwise.

bool lti::classifier2DVisualizer::apply	(	const supervisedInstanceClassifier &	cls,
		const lti::dmatrix &	highlights,
		image &	result
	)			const

Applies the given classifier to the area defined in the parameters object.

The image will contain the visualization result. It also highlights the vectors given in highlights.

Parameters:

	cls	supervisedInstanceClassifier with the classifier to be visualized. The classifier must be able to classify a 2D feature set.
	highlights	matrix with each row containing a vector that is to be highlighted.
	result	will receive the visualization result.

Returns:

true if apply successful or false otherwise.

bool lti::classifier2DVisualizer::apply	(	const supervisedInstanceClassifier &	cls,
		image &	result
	)			const

Applies the given classifier to the area defined in the parameters object.

The image will contain the visualization result.

Parameters:

	cls	supervisedInstanceClassifier with the classifier to be visualized. The classifier must be able to classify a 2D feature set.
	result	will receive the visualization result.

Returns:

true if apply successful or false otherwise.

virtual functor* lti::classifier2DVisualizer::clone

(

)

const [virtual]

returns a pointer to a clone of this functor.

Implements lti::functor.

classifier2DVisualizer& lti::classifier2DVisualizer::copy

(

const classifier2DVisualizer &

other

)

copy data of "other" functor.

Parameters:

other

the functor to be copied

Returns:

a reference to this functor object

Reimplemented from lti::functor.

const parameters& lti::classifier2DVisualizer::getParameters

(

)

const

returns used parameters

Reimplemented from lti::functor.

virtual const char* lti::classifier2DVisualizer::getTypeName

(

)

const [virtual]

returns the name of this type ("classifier2DVisualizer")

Reimplemented from lti::functor.

classifier2DVisualizer& lti::classifier2DVisualizer::operator=

(

const classifier2DVisualizer &

other

)

alias for copy member

Parameters:

other

the functor to be copied

Returns:

a reference to this functor object

Reimplemented from lti::functor.

---

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

• ltiClassifier2DVisualizer.h