third_party/opencv
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Merge pull request #3032 from vpisarev:refactor_ml2

Browse Source
This commit is contained in:
Vadim Pisarevsky
2014-08-04 14:51:37 +00:00
parent dbedc63c55 cb9b88e65b
commit 432b7cb235
68 changed files with 20952 additions and 32884 deletions
Download Patch File Download Diff File Expand all files Collapse all files
samples/cpp/agaricus-lepiota.data
-8124
View File
File diff suppressed because it is too large Load Diff
samples/cpp/bagofwords_classification.cpp
+63 -58
View File
@@ -23,6 +23,7 @@
#define DEBUG_DESC_PROGRESS
using namespace cv;
using namespace cv::ml;
using namespace std;
const string paramsFile = "params.xml";
@@ -677,7 +678,7 @@ void VocData::writeClassifierResultsFile( const string& out_dir, const string& o
result_file.close();
} else {
string err_msg = "could not open classifier results file '" + output_file + "' for writing. Before running for the first time, a 'results' subdirectory should be created within the VOC dataset base directory. e.g. if the VOC data is stored in /VOC/VOC2010 then the path /VOC/results must be created.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
}
@@ -701,9 +702,9 @@ void VocData::writeClassifierResultsFile( const string& out_dir, const string& o
string VocData::getResultsFilename(const string& obj_class, const VocTask task, const ObdDatasetType dataset, const int competition, const int number)
{
if ((competition < 1) && (competition != -1))
CV_Error(CV_StsBadArg,"competition argument should be a positive non-zero number or -1 to accept the default");
CV_Error(Error::StsBadArg,"competition argument should be a positive non-zero number or -1 to accept the default");
if ((number < 1) && (number != -1))
CV_Error(CV_StsBadArg,"number argument should be a positive non-zero number or -1 to accept the default");
CV_Error(Error::StsBadArg,"number argument should be a positive non-zero number or -1 to accept the default");
string dset, task_type;
@@ -815,7 +816,7 @@ void VocData::calcClassifierPrecRecall(const string& input_file, vector<float>&
scoregt_file.close();
} else {
string err_msg = "could not open scoregt file '" + scoregt_file_str + "' for writing.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
}
@@ -974,7 +975,7 @@ void VocData::calcClassifierConfMatRow(const string& obj_class, const vector<Obd
if (target_idx_it == output_headers.end())
{
string err_msg = "could not find the target object class '" + obj_class + "' in list of valid classes.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
/* convert iterator to index */
target_idx = (int)std::distance(output_headers.begin(),target_idx_it);
@@ -1037,7 +1038,7 @@ void VocData::calcClassifierConfMatRow(const string& obj_class, const vector<Obd
if (class_idx_it == output_headers.end())
{
string err_msg = "could not find object class '" + img_objects[obj_idx].object_class + "' specified in the ground truth file of '" + images[ranking[image_idx]].id +"'in list of valid classes.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
/* convert iterator to index */
int class_idx = (int)std::distance(output_headers.begin(),class_idx_it);
@@ -1189,7 +1190,7 @@ void VocData::calcDetectorConfMatRow(const string& obj_class, const ObdDatasetTy
if (class_idx_it == output_headers.end())
{
string err_msg = "could not find object class '" + img_objects[max_gt_obj_idx].object_class + "' specified in the ground truth file of '" + images[ranking[image_idx]].id +"'in list of valid classes.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
/* convert iterator to index */
int class_idx = (int)std::distance(output_headers.begin(),class_idx_it);
@@ -1282,7 +1283,7 @@ void VocData::savePrecRecallToGnuplot(const string& output_file, const vector<fl
plot_file.close();
} else {
string err_msg = "could not open plot file '" + output_file_std + "' for writing.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
}
@@ -1446,7 +1447,7 @@ void VocData::readClassifierGroundTruth(const string& filename, vector<string>&
if (!gtfile.is_open())
{
string err_msg = "could not open VOC ground truth textfile '" + filename + "'.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
string line;
@@ -1462,7 +1463,7 @@ void VocData::readClassifierGroundTruth(const string& filename, vector<string>&
image_codes.push_back(image);
object_present.push_back(obj_present == 1);
} else {
if (!gtfile.eof()) CV_Error(CV_StsParseError,"error parsing VOC ground truth textfile.");
if (!gtfile.eof()) CV_Error(Error::StsParseError,"error parsing VOC ground truth textfile.");
}
}
gtfile.close();
@@ -1488,13 +1489,13 @@ void VocData::readClassifierResultsFile(const string& input_file, vector<string>
image_codes.push_back(image);
scores.push_back(score);
} else {
if(!result_file.eof()) CV_Error(CV_StsParseError,"error parsing VOC classifier results file.");
if(!result_file.eof()) CV_Error(Error::StsParseError,"error parsing VOC classifier results file.");
}
}
result_file.close();
} else {
string err_msg = "could not open classifier results file '" + input_file + "' for reading.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
}
@@ -1545,13 +1546,13 @@ void VocData::readDetectorResultsFile(const string& input_file, vector<string>&
bounding_boxes[image_idx].push_back(bounding_box);
}
} else {
if(!result_file.eof()) CV_Error(CV_StsParseError,"error parsing VOC detector results file.");
if(!result_file.eof()) CV_Error(Error::StsParseError,"error parsing VOC detector results file.");
}
}
result_file.close();
} else {
string err_msg = "could not open detector results file '" + input_file + "' for reading.";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
}
@@ -1595,23 +1596,23 @@ void VocData::extractVocObjects(const string filename, vector<ObdObject>& object
//object class -------------
if (extractXMLBlock(object_contents, "name", 0, tag_contents) == -1) CV_Error(CV_StsError,"missing <name> tag in object definition of '" + filename + "'");
if (extractXMLBlock(object_contents, "name", 0, tag_contents) == -1) CV_Error(Error::StsError,"missing <name> tag in object definition of '" + filename + "'");
object.object_class.swap(tag_contents);
//object bounding box -------------
int xmax, xmin, ymax, ymin;
if (extractXMLBlock(object_contents, "xmax", 0, tag_contents) == -1) CV_Error(CV_StsError,"missing <xmax> tag in object definition of '" + filename + "'");
if (extractXMLBlock(object_contents, "xmax", 0, tag_contents) == -1) CV_Error(Error::StsError,"missing <xmax> tag in object definition of '" + filename + "'");
xmax = stringToInteger(tag_contents);
if (extractXMLBlock(object_contents, "xmin", 0, tag_contents) == -1) CV_Error(CV_StsError,"missing <xmin> tag in object definition of '" + filename + "'");
if (extractXMLBlock(object_contents, "xmin", 0, tag_contents) == -1) CV_Error(Error::StsError,"missing <xmin> tag in object definition of '" + filename + "'");
xmin = stringToInteger(tag_contents);
if (extractXMLBlock(object_contents, "ymax", 0, tag_contents) == -1) CV_Error(CV_StsError,"missing <ymax> tag in object definition of '" + filename + "'");
if (extractXMLBlock(object_contents, "ymax", 0, tag_contents) == -1) CV_Error(Error::StsError,"missing <ymax> tag in object definition of '" + filename + "'");
ymax = stringToInteger(tag_contents);
if (extractXMLBlock(object_contents, "ymin", 0, tag_contents) == -1) CV_Error(CV_StsError,"missing <ymin> tag in object definition of '" + filename + "'");
if (extractXMLBlock(object_contents, "ymin", 0, tag_contents) == -1) CV_Error(Error::StsError,"missing <ymin> tag in object definition of '" + filename + "'");
ymin = stringToInteger(tag_contents);
object.boundingBox.x = xmin-1; //convert to 0-based indexing
@@ -1714,11 +1715,11 @@ void VocData::extractDataFromResultsFilename(const string& input_file, string& c
size_t fnameend = input_file_std.rfind(".txt");
if ((fnamestart == input_file_std.npos) || (fnameend == input_file_std.npos))
CV_Error(CV_StsError,"Could not extract filename of results file.");
CV_Error(Error::StsError,"Could not extract filename of results file.");
++fnamestart;
if (fnamestart >= fnameend)
CV_Error(CV_StsError,"Could not extract filename of results file.");
CV_Error(Error::StsError,"Could not extract filename of results file.");
//extract dataset and class names, triggering exception if the filename format is not correct
string filename = input_file_std.substr(fnamestart, fnameend-fnamestart);
@@ -1729,11 +1730,11 @@ void VocData::extractDataFromResultsFilename(const string& input_file, string& c
size_t classend = filename.find("_",classstart+1);
if (classend == filename.npos) classend = filename.size();
if ((datasetstart == filename.npos) || (classstart == filename.npos))
CV_Error(CV_StsError,"Error parsing results filename. Is it in standard format of 'comp<n>_{cls/det}_<dataset>_<objclass>.txt'?");
CV_Error(Error::StsError,"Error parsing results filename. Is it in standard format of 'comp<n>_{cls/det}_<dataset>_<objclass>.txt'?");
++datasetstart;
++classstart;
if (((datasetstart-classstart) < 1) || ((classend-datasetstart) < 1))
CV_Error(CV_StsError,"Error parsing results filename. Is it in standard format of 'comp<n>_{cls/det}_<dataset>_<objclass>.txt'?");
CV_Error(Error::StsError,"Error parsing results filename. Is it in standard format of 'comp<n>_{cls/det}_<dataset>_<objclass>.txt'?");
dataset_name = filename.substr(datasetstart,classstart-datasetstart-1);
class_name = filename.substr(classstart,classend-classstart);
@@ -1781,7 +1782,7 @@ bool VocData::getClassifierGroundTruthImage(const string& obj_class, const strin
return m_classifier_gt_all_present[std::distance(m_classifier_gt_all_ids.begin(),it)] != 0;
} else {
string err_msg = "could not find classifier ground truth for image '" + id + "' and class '" + obj_class + "'";
CV_Error(CV_StsError,err_msg.c_str());
CV_Error(Error::StsError,err_msg.c_str());
}
return true;
@@ -1814,7 +1815,7 @@ void VocData::getSortOrder(const vector<float>& values, vector<size_t>& order, b
void VocData::readFileToString(const string filename, string& file_contents)
{
std::ifstream ifs(filename.c_str());
if (!ifs.is_open()) CV_Error(CV_StsError,"could not open text file");
if (!ifs.is_open()) CV_Error(Error::StsError,"could not open text file");
stringstream oss;
oss << ifs.rdbuf();
@@ -1829,7 +1830,7 @@ int VocData::stringToInteger(const string input_str)
stringstream ss(input_str);
if ((ss >> result).fail())
{
CV_Error(CV_StsBadArg,"could not perform string to integer conversion");
CV_Error(Error::StsBadArg,"could not perform string to integer conversion");
}
return result;
}
@@ -1841,7 +1842,7 @@ string VocData::integerToString(const int input_int)
stringstream ss;
if ((ss << input_int).fail())
{
CV_Error(CV_StsBadArg,"could not perform integer to string conversion");
CV_Error(Error::StsBadArg,"could not perform integer to string conversion");
}
result = ss.str();
return result;
@@ -2325,14 +2326,14 @@ static void removeBowImageDescriptorsByCount( vector<ObdImage>& images, vector<M
CV_Assert( bowImageDescriptors.size() == objectPresent.size() );
}
static void setSVMParams( CvSVMParams& svmParams, CvMat& class_wts_cv, const Mat& responses, bool balanceClasses )
static void setSVMParams( SVM::Params& svmParams, Mat& class_wts_cv, const Mat& responses, bool balanceClasses )
{
int pos_ex = countNonZero(responses == 1);
int neg_ex = countNonZero(responses == -1);
cout << pos_ex << " positive training samples; " << neg_ex << " negative training samples" << endl;
svmParams.svm_type = CvSVM::C_SVC;
svmParams.kernel_type = CvSVM::RBF;
svmParams.svmType = SVM::C_SVC;
svmParams.kernelType = SVM::RBF;
if( balanceClasses )
{
Mat class_wts( 2, 1, CV_32FC1 );
@@ -2350,43 +2351,44 @@ static void setSVMParams( CvSVMParams& svmParams, CvMat& class_wts_cv, const Mat
class_wts.at<float>(1) = static_cast<float>(pos_ex)/static_cast<float>(pos_ex+neg_ex);
}
class_wts_cv = class_wts;
svmParams.class_weights = &class_wts_cv;
svmParams.classWeights = class_wts_cv;
}
}
static void setSVMTrainAutoParams( CvParamGrid& c_grid, CvParamGrid& gamma_grid,
CvParamGrid& p_grid, CvParamGrid& nu_grid,
CvParamGrid& coef_grid, CvParamGrid& degree_grid )
static void setSVMTrainAutoParams( ParamGrid& c_grid, ParamGrid& gamma_grid,
ParamGrid& p_grid, ParamGrid& nu_grid,
ParamGrid& coef_grid, ParamGrid& degree_grid )
{
c_grid = CvSVM::get_default_grid(CvSVM::C);
c_grid = SVM::getDefaultGrid(SVM::C);
gamma_grid = CvSVM::get_default_grid(CvSVM::GAMMA);
gamma_grid = SVM::getDefaultGrid(SVM::GAMMA);
p_grid = CvSVM::get_default_grid(CvSVM::P);
p_grid.step = 0;
p_grid = SVM::getDefaultGrid(SVM::P);
p_grid.logStep = 0;
nu_grid = CvSVM::get_default_grid(CvSVM::NU);
nu_grid.step = 0;
nu_grid = SVM::getDefaultGrid(SVM::NU);
nu_grid.logStep = 0;
coef_grid = CvSVM::get_default_grid(CvSVM::COEF);
coef_grid.step = 0;
coef_grid = SVM::getDefaultGrid(SVM::COEF);
coef_grid.logStep = 0;
degree_grid = CvSVM::get_default_grid(CvSVM::DEGREE);
degree_grid.step = 0;
degree_grid = SVM::getDefaultGrid(SVM::DEGREE);
degree_grid.logStep = 0;
}
static void trainSVMClassifier( CvSVM& svm, const SVMTrainParamsExt& svmParamsExt, const string& objClassName, VocData& vocData,
static Ptr<SVM> trainSVMClassifier( const SVMTrainParamsExt& svmParamsExt, const string& objClassName, VocData& vocData,
Ptr<BOWImgDescriptorExtractor>& bowExtractor, const Ptr<FeatureDetector>& fdetector,
const string& resPath )
{
/* first check if a previously trained svm for the current class has been saved to file */
string svmFilename = resPath + svmsDir + "/" + objClassName + ".xml.gz";
Ptr<SVM> svm;
FileStorage fs( svmFilename, FileStorage::READ);
if( fs.isOpened() )
{
cout << "*** LOADING SVM CLASSIFIER FOR CLASS " << objClassName << " ***" << endl;
svm.load( svmFilename.c_str() );
svm = StatModel::load<SVM>( svmFilename );
}
else
{
@@ -2437,20 +2439,24 @@ static void trainSVMClassifier( CvSVM& svm, const SVMTrainParamsExt& svmParamsEx
}
cout << "TRAINING SVM FOR CLASS ..." << objClassName << "..." << endl;
CvSVMParams svmParams;
CvMat class_wts_cv;
SVM::Params svmParams;
Mat class_wts_cv;
setSVMParams( svmParams, class_wts_cv, responses, svmParamsExt.balanceClasses );
CvParamGrid c_grid, gamma_grid, p_grid, nu_grid, coef_grid, degree_grid;
svm = SVM::create(svmParams);
ParamGrid c_grid, gamma_grid, p_grid, nu_grid, coef_grid, degree_grid;
setSVMTrainAutoParams( c_grid, gamma_grid, p_grid, nu_grid, coef_grid, degree_grid );
svm.train_auto( trainData, responses, Mat(), Mat(), svmParams, 10, c_grid, gamma_grid, p_grid, nu_grid, coef_grid, degree_grid );
svm->trainAuto(TrainData::create(trainData, ROW_SAMPLE, responses), 10,
c_grid, gamma_grid, p_grid, nu_grid, coef_grid, degree_grid);
cout << "SVM TRAINING FOR CLASS " << objClassName << " COMPLETED" << endl;
svm.save( svmFilename.c_str() );
svm->save( svmFilename );
cout << "SAVED CLASSIFIER TO FILE" << endl;
}
return svm;
}
static void computeConfidences( CvSVM& svm, const string& objClassName, VocData& vocData,
static void computeConfidences( const Ptr<SVM>& svm, const string& objClassName, VocData& vocData,
Ptr<BOWImgDescriptorExtractor>& bowExtractor, const Ptr<FeatureDetector>& fdetector,
const string& resPath )
{
@@ -2476,12 +2482,12 @@ static void computeConfidences( CvSVM& svm, const string& objClassName, VocData&
if( imageIdx == 0 )
{
// In the first iteration, determine the sign of the positive class
float classVal = confidences[imageIdx] = svm.predict( bowImageDescriptors[imageIdx], false );
float scoreVal = confidences[imageIdx] = svm.predict( bowImageDescriptors[imageIdx], true );
float classVal = confidences[imageIdx] = svm->predict( bowImageDescriptors[imageIdx], noArray(), 0 );
float scoreVal = confidences[imageIdx] = svm->predict( bowImageDescriptors[imageIdx], noArray(), StatModel::RAW_OUTPUT );
signMul = (classVal < 0) == (scoreVal < 0) ? 1.f : -1.f;
}
// svm output of decision function
confidences[imageIdx] = signMul * svm.predict( bowImageDescriptors[imageIdx], true );
confidences[imageIdx] = signMul * svm->predict( bowImageDescriptors[imageIdx], noArray(), StatModel::RAW_OUTPUT );
}
cout << "WRITING QUERY RESULTS TO VOC RESULTS FILE FOR CLASS " << objClassName << "..." << endl;
@@ -2591,9 +2597,8 @@ int main(int argc, char** argv)
for( size_t classIdx = 0; classIdx < objClasses.size(); ++classIdx )
{
// Train a classifier on train dataset
CvSVM svm;
trainSVMClassifier( svm, svmTrainParamsExt, objClasses[classIdx], vocData,
bowExtractor, featureDetector, resPath );
Ptr<SVM> svm = trainSVMClassifier( svmTrainParamsExt, objClasses[classIdx], vocData,
bowExtractor, featureDetector, resPath );
// Now use the classifier over all images on the test dataset and rank according to score order
// also calculating precision-recall etc.
samples/cpp/em.cpp
+5 -3
View File
@@ -2,6 +2,7 @@
#include "opencv2/ml.hpp"
using namespace cv;
using namespace cv::ml;
int main( int /*argc*/, char** /*argv*/ )
{
@@ -34,8 +35,9 @@ int main( int /*argc*/, char** /*argv*/ )
samples = samples.reshape(1, 0);
// cluster the data
EM em_model(N, EM::COV_MAT_SPHERICAL, TermCriteria(TermCriteria::COUNT+TermCriteria::EPS, 300, 0.1));
em_model.train( samples, noArray(), labels, noArray() );
Ptr<EM> em_model = EM::train( samples, noArray(), labels, noArray(),
EM::Params(N, EM::COV_MAT_SPHERICAL,
TermCriteria(TermCriteria::COUNT+TermCriteria::EPS, 300, 0.1)));
// classify every image pixel
for( i = 0; i < img.rows; i++ )
@@ -44,7 +46,7 @@ int main( int /*argc*/, char** /*argv*/ )
{
sample.at<float>(0) = (float)j;
sample.at<float>(1) = (float)i;
int response = cvRound(em_model.predict( sample )[1]);
int response = cvRound(em_model->predict2( sample, noArray() )[1]);
Scalar c = colors[response];
circle( img, Point(j, i), 1, c*0.75, FILLED );
samples/cpp/letter_recog.cpp
+285 -534
View File
File diff suppressed because it is too large Load Diff
samples/cpp/mushroom.cpp
-322
View File
@@ -1,322 +0,0 @@
#include "opencv2/core/core_c.h"
#include "opencv2/ml/ml.hpp"
#include <stdio.h>
static void help()
{
printf("\nThis program demonstrated the use of OpenCV's decision tree function for learning and predicting data\n"
"Usage :\n"
"./mushroom <path to agaricus-lepiota.data>\n"
"\n"
"The sample demonstrates how to build a decision tree for classifying mushrooms.\n"
"It uses the sample base agaricus-lepiota.data from UCI Repository, here is the link:\n"
"\n"
"Newman, D.J. & Hettich, S. & Blake, C.L. & Merz, C.J. (1998).\n"
"UCI Repository of machine learning databases\n"
"[http://www.ics.uci.edu/~mlearn/MLRepository.html].\n"
"Irvine, CA: University of California, Department of Information and Computer Science.\n"
"\n"
"// loads the mushroom database, which is a text file, containing\n"
"// one training sample per row, all the input variables and the output variable are categorical,\n"
"// the values are encoded by characters.\n\n");
}
static int mushroom_read_database( const char* filename, CvMat** data, CvMat** missing, CvMat** responses )
{
const int M = 1024;
FILE* f = fopen( filename, "rt" );
CvMemStorage* storage;
CvSeq* seq;
char buf[M+2], *ptr;
float* el_ptr;
CvSeqReader reader;
int i, j, var_count = 0;
if( !f )
return 0;
// read the first line and determine the number of variables
if( !fgets( buf, M, f ))
{
fclose(f);
return 0;
}
for( ptr = buf; *ptr != '\0'; ptr++ )
var_count += *ptr == ',';
assert( ptr - buf == (var_count+1)*2 );
// create temporary memory storage to store the whole database
el_ptr = new float[var_count+1];
storage = cvCreateMemStorage();
seq = cvCreateSeq( 0, sizeof(*seq), (var_count+1)*sizeof(float), storage );
for(;;)
{
for( i = 0; i <= var_count; i++ )
{
int c = buf[i*2];
el_ptr[i] = c == '?' ? -1.f : (float)c;
}
if( i != var_count+1 )
break;
cvSeqPush( seq, el_ptr );
if( !fgets( buf, M, f ) || !strchr( buf, ',' ) )
break;
}
fclose(f);
// allocate the output matrices and copy the base there
*data = cvCreateMat( seq->total, var_count, CV_32F );
*missing = cvCreateMat( seq->total, var_count, CV_8U );
*responses = cvCreateMat( seq->total, 1, CV_32F );
cvStartReadSeq( seq, &reader );
for( i = 0; i < seq->total; i++ )
{
const float* sdata = (float*)reader.ptr + 1;
float* ddata = data[0]->data.fl + var_count*i;
float* dr = responses[0]->data.fl + i;
uchar* dm = missing[0]->data.ptr + var_count*i;
for( j = 0; j < var_count; j++ )
{
ddata[j] = sdata[j];
dm[j] = sdata[j] < 0;
}
*dr = sdata[-1];
CV_NEXT_SEQ_ELEM( seq->elem_size, reader );
}
cvReleaseMemStorage( &storage );
delete [] el_ptr;
return 1;
}
static CvDTree* mushroom_create_dtree( const CvMat* data, const CvMat* missing,
const CvMat* responses, float p_weight )
{
CvDTree* dtree;
CvMat* var_type;
int i, hr1 = 0, hr2 = 0, p_total = 0;
float priors[] = { 1, p_weight };
var_type = cvCreateMat( data->cols + 1, 1, CV_8U );
cvSet( var_type, cvScalarAll(CV_VAR_CATEGORICAL) ); // all the variables are categorical
dtree = new CvDTree;
dtree->train( data, CV_ROW_SAMPLE, responses, 0, 0, var_type, missing,
CvDTreeParams( 8, // max depth
10, // min sample count
0, // regression accuracy: N/A here
true, // compute surrogate split, as we have missing data
15, // max number of categories (use sub-optimal algorithm for larger numbers)
10, // the number of cross-validation folds
true, // use 1SE rule => smaller tree
true, // throw away the pruned tree branches
priors // the array of priors, the bigger p_weight, the more attention
// to the poisonous mushrooms
// (a mushroom will be judjed to be poisonous with bigger chance)
));
// compute hit-rate on the training database, demonstrates predict usage.
for( i = 0; i < data->rows; i++ )
{
CvMat sample, mask;
cvGetRow( data, &sample, i );
cvGetRow( missing, &mask, i );
double r = dtree->predict( &sample, &mask )->value;
int d = fabs(r - responses->data.fl[i]) >= FLT_EPSILON;
if( d )
{
if( r != 'p' )
hr1++;
else
hr2++;
}
p_total += responses->data.fl[i] == 'p';
}
printf( "Results on the training database:\n"
"\tPoisonous mushrooms mis-predicted: %d (%g%%)\n"
"\tFalse-alarms: %d (%g%%)\n", hr1, (double)hr1*100/p_total,
hr2, (double)hr2*100/(data->rows - p_total) );
cvReleaseMat( &var_type );
return dtree;
}
static const char* var_desc[] =
{
"cap shape (bell=b,conical=c,convex=x,flat=f)",
"cap surface (fibrous=f,grooves=g,scaly=y,smooth=s)",
"cap color (brown=n,buff=b,cinnamon=c,gray=g,green=r,\n\tpink=p,purple=u,red=e,white=w,yellow=y)",
"bruises? (bruises=t,no=f)",
"odor (almond=a,anise=l,creosote=c,fishy=y,foul=f,\n\tmusty=m,none=n,pungent=p,spicy=s)",
"gill attachment (attached=a,descending=d,free=f,notched=n)",
"gill spacing (close=c,crowded=w,distant=d)",
"gill size (broad=b,narrow=n)",
"gill color (black=k,brown=n,buff=b,chocolate=h,gray=g,\n\tgreen=r,orange=o,pink=p,purple=u,red=e,white=w,yellow=y)",
"stalk shape (enlarging=e,tapering=t)",
"stalk root (bulbous=b,club=c,cup=u,equal=e,rhizomorphs=z,rooted=r)",
"stalk surface above ring (ibrous=f,scaly=y,silky=k,smooth=s)",
"stalk surface below ring (ibrous=f,scaly=y,silky=k,smooth=s)",
"stalk color above ring (brown=n,buff=b,cinnamon=c,gray=g,orange=o,\n\tpink=p,red=e,white=w,yellow=y)",
"stalk color below ring (brown=n,buff=b,cinnamon=c,gray=g,orange=o,\n\tpink=p,red=e,white=w,yellow=y)",
"veil type (partial=p,universal=u)",
"veil color (brown=n,orange=o,white=w,yellow=y)",
"ring number (none=n,one=o,two=t)",
"ring type (cobwebby=c,evanescent=e,flaring=f,large=l,\n\tnone=n,pendant=p,sheathing=s,zone=z)",
"spore print color (black=k,brown=n,buff=b,chocolate=h,green=r,\n\torange=o,purple=u,white=w,yellow=y)",
"population (abundant=a,clustered=c,numerous=n,\n\tscattered=s,several=v,solitary=y)",
"habitat (grasses=g,leaves=l,meadows=m,paths=p\n\turban=u,waste=w,woods=d)",
0
};
static void print_variable_importance( CvDTree* dtree )
{
const CvMat* var_importance = dtree->get_var_importance();
int i;
char input[1000];
if( !var_importance )
{
printf( "Error: Variable importance can not be retrieved\n" );
return;
}
printf( "Print variable importance information? (y/n) " );
int values_read = scanf( "%1s", input );
CV_Assert(values_read == 1);
if( input[0] != 'y' && input[0] != 'Y' )
return;
for( i = 0; i < var_importance->cols*var_importance->rows; i++ )
{
double val = var_importance->data.db[i];
char buf[100];
int len = (int)(strchr( var_desc[i], '(' ) - var_desc[i] - 1);
strncpy( buf, var_desc[i], len );
buf[len] = '\0';
printf( "%s", buf );
printf( ": %g%%\n", val*100. );
}
}
static void interactive_classification( CvDTree* dtree )
{
char input[1000];
const CvDTreeNode* root;
CvDTreeTrainData* data;
if( !dtree )
return;
root = dtree->get_root();
data = dtree->get_data();
for(;;)
{
const CvDTreeNode* node;
printf( "Start/Proceed with interactive mushroom classification (y/n): " );
int values_read = scanf( "%1s", input );
CV_Assert(values_read == 1);
if( input[0] != 'y' && input[0] != 'Y' )
break;
printf( "Enter 1-letter answers, '?' for missing/unknown value...\n" );
// custom version of predict
node = root;
for(;;)
{
CvDTreeSplit* split = node->split;
int dir = 0;
if( !node->left || node->Tn <= dtree->get_pruned_tree_idx() || !node->split )
break;
for( ; split != 0; )
{
int vi = split->var_idx, j;
int count = data->cat_count->data.i[vi];
const int* map = data->cat_map->data.i + data->cat_ofs->data.i[vi];
printf( "%s: ", var_desc[vi] );
values_read = scanf( "%1s", input );
CV_Assert(values_read == 1);
if( input[0] == '?' )
{
split = split->next;
continue;
}
// convert the input character to the normalized value of the variable
for( j = 0; j < count; j++ )
if( map[j] == input[0] )
break;
if( j < count )
{
dir = (split->subset[j>>5] & (1 << (j&31))) ? -1 : 1;
if( split->inversed )
dir = -dir;
break;
}
else
printf( "Error: unrecognized value\n" );
}
if( !dir )
{
printf( "Impossible to classify the sample\n");
node = 0;
break;
}
node = dir < 0 ? node->left : node->right;
}
if( node )
printf( "Prediction result: the mushroom is %s\n",
node->class_idx == 0 ? "EDIBLE" : "POISONOUS" );
printf( "\n-----------------------------\n" );
}
}
int main( int argc, char** argv )
{
CvMat *data = 0, *missing = 0, *responses = 0;
CvDTree* dtree;
const char* base_path = argc >= 2 ? argv[1] : "agaricus-lepiota.data";
help();
if( !mushroom_read_database( base_path, &data, &missing, &responses ) )
{
printf( "\nUnable to load the training database\n\n");
help();
return -1;
}
dtree = mushroom_create_dtree( data, missing, responses,
10 // poisonous mushrooms will have 10x higher weight in the decision tree
);
cvReleaseMat( &data );
cvReleaseMat( &missing );
cvReleaseMat( &responses );
print_variable_importance( dtree );
interactive_classification( dtree );
delete dtree;
return 0;
}
samples/cpp/points_classifier.cpp
+137 -378
View File
@@ -12,6 +12,7 @@
using namespace std;
using namespace cv;
using namespace cv::ml;
const Scalar WHITE_COLOR = Scalar(255,255,255);
const string winName = "points";
@@ -22,18 +23,20 @@ RNG rng;
vector<Point> trainedPoints;
vector<int> trainedPointsMarkers;
vector<Scalar> classColors;
const int MAX_CLASSES = 2;
vector<Vec3b> classColors(MAX_CLASSES);
int currentClass = 0;
vector<int> classCounters(MAX_CLASSES);
#define _NBC_ 0 // normal Bayessian classifier
#define _KNN_ 0 // k nearest neighbors classifier
#define _SVM_ 0 // support vectors machine
#define _NBC_ 1 // normal Bayessian classifier
#define _KNN_ 1 // k nearest neighbors classifier
#define _SVM_ 1 // support vectors machine
#define _DT_ 1 // decision tree
#define _BT_ 0 // ADA Boost
#define _BT_ 1 // ADA Boost
#define _GBT_ 0 // gradient boosted trees
#define _RF_ 0 // random forest
#define _ERT_ 0 // extremely randomized trees
#define _ANN_ 0 // artificial neural networks
#define _EM_ 0 // expectation-maximization
#define _RF_ 1 // random forest
#define _ANN_ 1 // artificial neural networks
#define _EM_ 1 // expectation-maximization
static void on_mouse( int event, int x, int y, int /*flags*/, void* )
{
@@ -44,76 +47,43 @@ static void on_mouse( int event, int x, int y, int /*flags*/, void* )
if( event == EVENT_LBUTTONUP )
{
if( classColors.empty() )
return;
trainedPoints.push_back( Point(x,y) );
trainedPointsMarkers.push_back( (int)(classColors.size()-1) );
trainedPointsMarkers.push_back( currentClass );
classCounters[currentClass]++;
updateFlag = true;
}
else if( event == EVENT_RBUTTONUP )
{
#if _BT_
if( classColors.size() < 2 )
{
#endif
classColors.push_back( Scalar((uchar)rng(256), (uchar)rng(256), (uchar)rng(256)) );
updateFlag = true;
#if _BT_
}
else
cout << "New class can not be added, because CvBoost can only be used for 2-class classification" << endl;
#endif
}
//draw
if( updateFlag )
{
img = Scalar::all(0);
// put the text
stringstream text;
text << "current class " << classColors.size()-1;
putText( img, text.str(), Point(10,25), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
text.str("");
text << "total classes " << classColors.size();
putText( img, text.str(), Point(10,50), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
text.str("");
text << "total points " << trainedPoints.size();
putText(img, text.str(), Point(10,75), FONT_HERSHEY_SIMPLEX, 0.8f, WHITE_COLOR, 2 );
// draw points
for( size_t i = 0; i < trainedPoints.size(); i++ )
circle( img, trainedPoints[i], 5, classColors[trainedPointsMarkers[i]], -1 );
{
Vec3b c = classColors[trainedPointsMarkers[i]];
circle( img, trainedPoints[i], 5, Scalar(c), -1 );
}
imshow( winName, img );
}
}
static void prepare_train_data( Mat& samples, Mat& classes )
static Mat prepare_train_samples(const vector<Point>& pts)
{
Mat( trainedPoints ).copyTo( samples );
Mat( trainedPointsMarkers ).copyTo( classes );
// reshape trainData and change its type
samples = samples.reshape( 1, samples.rows );
samples.convertTo( samples, CV_32FC1 );
Mat samples;
Mat(pts).reshape(1, (int)pts.size()).convertTo(samples, CV_32F);
return samples;
}
#if _NBC_
static void find_decision_boundary_NBC()
static Ptr<TrainData> prepare_train_data()
{
img.copyTo( imgDst );
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
CvNormalBayesClassifier normalBayesClassifier( trainSamples, trainClasses );
Mat samples = prepare_train_samples(trainedPoints);
return TrainData::create(samples, ROW_SAMPLE, Mat(trainedPointsMarkers));
}
static void predict_and_paint(const Ptr<StatModel>& model, Mat& dst)
{
Mat testSample( 1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
@@ -122,378 +92,171 @@ static void find_decision_boundary_NBC()
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)normalBayesClassifier.predict( testSample );
circle( imgDst, Point(x,y), 1, classColors[response] );
int response = (int)model->predict( testSample );
dst.at<Vec3b>(y, x) = classColors[response];
}
}
}
#if _NBC_
static void find_decision_boundary_NBC()
{
// learn classifier
Ptr<NormalBayesClassifier> normalBayesClassifier = StatModel::train<NormalBayesClassifier>(prepare_train_data(), NormalBayesClassifier::Params());
predict_and_paint(normalBayesClassifier, imgDst);
}
#endif
#if _KNN_
static void find_decision_boundary_KNN( int K )
{
img.copyTo( imgDst );
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
#if defined HAVE_OPENCV_OCL && _OCL_KNN_
cv::ocl::KNearestNeighbour knnClassifier;
Mat temp, result;
knnClassifier.train(trainSamples, trainClasses, temp, false, K);
cv::ocl::oclMat testSample_ocl, reslut_ocl;
#else
CvKNearest knnClassifier( trainSamples, trainClasses, Mat(), false, K );
#endif
Mat testSample( 1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
#if defined HAVE_OPENCV_OCL && _OCL_KNN_
testSample_ocl.upload(testSample);
knnClassifier.find_nearest(testSample_ocl, K, reslut_ocl);
reslut_ocl.download(result);
int response = saturate_cast<int>(result.at<float>(0));
circle(imgDst, Point(x, y), 1, classColors[response]);
#else
int response = (int)knnClassifier.find_nearest( testSample, K );
circle( imgDst, Point(x,y), 1, classColors[response] );
#endif
}
}
Ptr<KNearest> knn = StatModel::train<KNearest>(prepare_train_data(), KNearest::Params(K, true));
predict_and_paint(knn, imgDst);
}
#endif
#if _SVM_
static void find_decision_boundary_SVM( CvSVMParams params )
static void find_decision_boundary_SVM( SVM::Params params )
{
img.copyTo( imgDst );
Ptr<SVM> svm = StatModel::train<SVM>(prepare_train_data(), params);
predict_and_paint(svm, imgDst);
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
#if defined HAVE_OPENCV_OCL && _OCL_SVM_
cv::ocl::CvSVM_OCL svmClassifier(trainSamples, trainClasses, Mat(), Mat(), params);
#else
CvSVM svmClassifier( trainSamples, trainClasses, Mat(), Mat(), params );
#endif
Mat testSample( 1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
Mat sv = svm->getSupportVectors();
for( int i = 0; i < sv.rows; i++ )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)svmClassifier.predict( testSample );
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
const float* supportVector = sv.ptr<float>(i);
circle( imgDst, Point(saturate_cast<int>(supportVector[0]),saturate_cast<int>(supportVector[1])), 5, Scalar(255,255,255), -1 );
}
for( int i = 0; i < svmClassifier.get_support_vector_count(); i++ )
{
const float* supportVector = svmClassifier.get_support_vector(i);
circle( imgDst, Point(saturate_cast<int>(supportVector[0]),saturate_cast<int>(supportVector[1])), 5, CV_RGB(255,255,255), -1 );
}
}
#endif
#if _DT_
static void find_decision_boundary_DT()
{
img.copyTo( imgDst );
DTrees::Params params;
params.maxDepth = 8;
params.minSampleCount = 2;
params.useSurrogates = false;
params.CVFolds = 0; // the number of cross-validation folds
params.use1SERule = false;
params.truncatePrunedTree = false;
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
Ptr<DTrees> dtree = StatModel::train<DTrees>(prepare_train_data(), params);
// learn classifier
CvDTree dtree;
Mat var_types( 1, trainSamples.cols + 1, CV_8UC1, Scalar(CV_VAR_ORDERED) );
var_types.at<uchar>( trainSamples.cols ) = CV_VAR_CATEGORICAL;
CvDTreeParams params;
params.max_depth = 8;
params.min_sample_count = 2;
params.use_surrogates = false;
params.cv_folds = 0; // the number of cross-validation folds
params.use_1se_rule = false;
params.truncate_pruned_tree = false;
dtree.train( trainSamples, CV_ROW_SAMPLE, trainClasses,
Mat(), Mat(), var_types, Mat(), params );
Mat testSample(1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)dtree.predict( testSample )->value;
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
}
predict_and_paint(dtree, imgDst);
}
#endif
#if _BT_
void find_decision_boundary_BT()
static void find_decision_boundary_BT()
{
img.copyTo( imgDst );
Boost::Params params( Boost::DISCRETE, // boost_type
100, // weak_count
0.95, // weight_trim_rate
2, // max_depth
false, //use_surrogates
Mat() // priors
);
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
CvBoost boost;
Mat var_types( 1, trainSamples.cols + 1, CV_8UC1, Scalar(CV_VAR_ORDERED) );
var_types.at<uchar>( trainSamples.cols ) = CV_VAR_CATEGORICAL;
CvBoostParams params( CvBoost::DISCRETE, // boost_type
100, // weak_count
0.95, // weight_trim_rate
2, // max_depth
false, //use_surrogates
0 // priors
);
boost.train( trainSamples, CV_ROW_SAMPLE, trainClasses, Mat(), Mat(), var_types, Mat(), params );
Mat testSample(1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)boost.predict( testSample );
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
}
Ptr<Boost> boost = StatModel::train<Boost>(prepare_train_data(), params);
predict_and_paint(boost, imgDst);
}
#endif
#if _GBT_
void find_decision_boundary_GBT()
static void find_decision_boundary_GBT()
{
img.copyTo( imgDst );
GBTrees::Params params( GBTrees::DEVIANCE_LOSS, // loss_function_type
100, // weak_count
0.1f, // shrinkage
1.0f, // subsample_portion
2, // max_depth
false // use_surrogates )
);
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
CvGBTrees gbtrees;
Mat var_types( 1, trainSamples.cols + 1, CV_8UC1, Scalar(CV_VAR_ORDERED) );
var_types.at<uchar>( trainSamples.cols ) = CV_VAR_CATEGORICAL;
CvGBTreesParams params( CvGBTrees::DEVIANCE_LOSS, // loss_function_type
100, // weak_count
0.1f, // shrinkage
1.0f, // subsample_portion
2, // max_depth
false // use_surrogates )
);
gbtrees.train( trainSamples, CV_ROW_SAMPLE, trainClasses, Mat(), Mat(), var_types, Mat(), params );
Mat testSample(1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)gbtrees.predict( testSample );
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
}
Ptr<GBTrees> gbtrees = StatModel::train<GBTrees>(prepare_train_data(), params);
predict_and_paint(gbtrees, imgDst);
}
#endif
#if _RF_
void find_decision_boundary_RF()
static void find_decision_boundary_RF()
{
img.copyTo( imgDst );
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
CvRTrees rtrees;
CvRTParams params( 4, // max_depth,
RTrees::Params params( 4, // max_depth,
2, // min_sample_count,
0.f, // regression_accuracy,
false, // use_surrogates,
16, // max_categories,
0, // priors,
Mat(), // priors,
false, // calc_var_importance,
1, // nactive_vars,
5, // max_num_of_trees_in_the_forest,
0, // forest_accuracy,
CV_TERMCRIT_ITER // termcrit_type
TermCriteria(TermCriteria::MAX_ITER, 5, 0) // max_num_of_trees_in_the_forest,
);
rtrees.train( trainSamples, CV_ROW_SAMPLE, trainClasses, Mat(), Mat(), Mat(), Mat(), params );
Mat testSample(1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)rtrees.predict( testSample );
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
}
Ptr<RTrees> rtrees = StatModel::train<RTrees>(prepare_train_data(), params);
predict_and_paint(rtrees, imgDst);
}
#endif
#if _ERT_
void find_decision_boundary_ERT()
{
img.copyTo( imgDst );
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// learn classifier
CvERTrees ertrees;
Mat var_types( 1, trainSamples.cols + 1, CV_8UC1, Scalar(CV_VAR_ORDERED) );
var_types.at<uchar>( trainSamples.cols ) = CV_VAR_CATEGORICAL;
CvRTParams params( 4, // max_depth,
2, // min_sample_count,
0.f, // regression_accuracy,
false, // use_surrogates,
16, // max_categories,
0, // priors,
false, // calc_var_importance,
1, // nactive_vars,
5, // max_num_of_trees_in_the_forest,
0, // forest_accuracy,
CV_TERMCRIT_ITER // termcrit_type
);
ertrees.train( trainSamples, CV_ROW_SAMPLE, trainClasses, Mat(), Mat(), var_types, Mat(), params );
Mat testSample(1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
int response = (int)ertrees.predict( testSample );
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
}
}
}
#endif
#if _ANN_
void find_decision_boundary_ANN( const Mat& layer_sizes )
static void find_decision_boundary_ANN( const Mat& layer_sizes )
{
img.copyTo( imgDst );
ANN_MLP::Params params(layer_sizes, ANN_MLP::SIGMOID_SYM, 1, 1, TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 300, FLT_EPSILON),
ANN_MLP::Params::BACKPROP, 0.001);
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
// prerare trainClasses
trainClasses.create( trainedPoints.size(), classColors.size(), CV_32FC1 );
for( int i = 0; i < trainClasses.rows; i++ )
Mat trainClasses = Mat::zeros( (int)trainedPoints.size(), (int)classColors.size(), CV_32FC1 );
for( int i = 0; i < trainClasses.rows; i++ )
{
for( int k = 0; k < trainClasses.cols; k++ )
{
if( k == trainedPointsMarkers[i] )
trainClasses.at<float>(i,k) = 1;
else
trainClasses.at<float>(i,k) = 0;
}
trainClasses.at<float>(i, trainedPointsMarkers[i]) = 1.f;
}
Mat weights( 1, trainedPoints.size(), CV_32FC1, Scalar::all(1) );
Mat samples = prepare_train_samples(trainedPoints);
Ptr<TrainData> tdata = TrainData::create(samples, ROW_SAMPLE, trainClasses);
// learn classifier
CvANN_MLP ann( layer_sizes, CvANN_MLP::SIGMOID_SYM, 1, 1 );
ann.train( trainSamples, trainClasses, weights );
Mat testSample( 1, 2, CV_32FC1 );
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
{
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
Mat outputs( 1, classColors.size(), CV_32FC1, testSample.data );
ann.predict( testSample, outputs );
Point maxLoc;
minMaxLoc( outputs, 0, 0, 0, &maxLoc );
circle( imgDst, Point(x,y), 2, classColors[maxLoc.x], 1 );
}
}
Ptr<ANN_MLP> ann = StatModel::train<ANN_MLP>(tdata, params);
predict_and_paint(ann, imgDst);
}
#endif
#if _EM_
void find_decision_boundary_EM()
static void find_decision_boundary_EM()
{
img.copyTo( imgDst );
Mat trainSamples, trainClasses;
prepare_train_data( trainSamples, trainClasses );
Mat samples = prepare_train_samples(trainedPoints);
vector<cv::EM> em_models(classColors.size());
int i, j, nmodels = (int)classColors.size();
vector<Ptr<EM> > em_models(nmodels);
Mat modelSamples;
CV_Assert((int)trainClasses.total() == trainSamples.rows);
CV_Assert((int)trainClasses.type() == CV_32SC1);
for(size_t modelIndex = 0; modelIndex < em_models.size(); modelIndex++)
for( i = 0; i < nmodels; i++ )
{
const int componentCount = 3;
em_models[modelIndex] = EM(componentCount, cv::EM::COV_MAT_DIAGONAL);
Mat modelSamples;
for(int sampleIndex = 0; sampleIndex < trainSamples.rows; sampleIndex++)
modelSamples.release();
for( j = 0; j < samples.rows; j++ )
{
if(trainClasses.at<int>(sampleIndex) == (int)modelIndex)
modelSamples.push_back(trainSamples.row(sampleIndex));
if( trainedPointsMarkers[j] == i )
modelSamples.push_back(samples.row(j));
}
// learn models
if(!modelSamples.empty())
em_models[modelIndex].train(modelSamples);
if( !modelSamples.empty() )
{
em_models[i] = EM::train(modelSamples, noArray(), noArray(), noArray(),
EM::Params(componentCount, EM::COV_MAT_DIAGONAL));
}
}
// classify coordinate plane points using the bayes classifier, i.e.
// y(x) = arg max_i=1_modelsCount likelihoods_i(x)
Mat testSample(1, 2, CV_32FC1 );
Mat logLikelihoods(1, nmodels, CV_64FC1, Scalar(-DBL_MAX));
for( int y = 0; y < img.rows; y += testStep )
{
for( int x = 0; x < img.cols; x += testStep )
@@ -501,17 +264,14 @@ void find_decision_boundary_EM()
testSample.at<float>(0) = (float)x;
testSample.at<float>(1) = (float)y;
Mat logLikelihoods(1, em_models.size(), CV_64FC1, Scalar(-DBL_MAX));
for(size_t modelIndex = 0; modelIndex < em_models.size(); modelIndex++)
for( i = 0; i < nmodels; i++ )
{
if(em_models[modelIndex].isTrained())
logLikelihoods.at<double>(modelIndex) = em_models[modelIndex].predict(testSample)[0];
if( !em_models[i].empty() )
logLikelihoods.at<double>(i) = em_models[i]->predict2(testSample, noArray())[0];
}
Point maxLoc;
minMaxLoc(logLikelihoods, 0, 0, 0, &maxLoc);
int response = maxLoc.x;
circle( imgDst, Point(x,y), 2, classColors[response], 1 );
imgDst.at<Vec3b>(y, x) = classColors[maxLoc.x];
}
}
}
@@ -520,7 +280,7 @@ void find_decision_boundary_EM()
int main()
{
cout << "Use:" << endl
<< " right mouse button - to add new class;" << endl
<< " key '0' .. '1' - switch to class #n" << endl
<< " left mouse button - to add new point;" << endl
<< " key 'r' - to run the ML model;" << endl
<< " key 'i' - to init (clear) the data." << endl << endl;
@@ -532,6 +292,9 @@ int main()
imshow( "points", img );
setMouseCallback( "points", on_mouse );
classColors[0] = Vec3b(0, 255, 0);
classColors[1] = Vec3b(0, 0, 255);
for(;;)
{
uchar key = (uchar)waitKey();
@@ -542,98 +305,94 @@ int main()
{
img = Scalar::all(0);
classColors.clear();
trainedPoints.clear();
trainedPointsMarkers.clear();
classCounters.assign(MAX_CLASSES, 0);
imshow( winName, img );
}
if( key == '0' || key == '1' )
{
currentClass = key - '0';
}
if( key == 'r' ) // run
{
double minVal = 0;
minMaxLoc(classCounters, &minVal, 0, 0, 0);
if( minVal == 0 )
{
printf("each class should have at least 1 point\n");
continue;
}
img.copyTo( imgDst );
#if _NBC_
find_decision_boundary_NBC();
namedWindow( "NormalBayesClassifier", WINDOW_AUTOSIZE );
imshow( "NormalBayesClassifier", imgDst );
#endif
#if _KNN_
int K = 3;
find_decision_boundary_KNN( K );
namedWindow( "kNN", WINDOW_AUTOSIZE );
imshow( "kNN", imgDst );
K = 15;
find_decision_boundary_KNN( K );
namedWindow( "kNN2", WINDOW_AUTOSIZE );
imshow( "kNN2", imgDst );
#endif
#if _SVM_
//(1)-(2)separable and not sets
CvSVMParams params;
params.svm_type = CvSVM::C_SVC;
params.kernel_type = CvSVM::POLY; //CvSVM::LINEAR;
SVM::Params params;
params.svmType = SVM::C_SVC;
params.kernelType = SVM::POLY; //CvSVM::LINEAR;
params.degree = 0.5;
params.gamma = 1;
params.coef0 = 1;
params.C = 1;
params.nu = 0.5;
params.p = 0;
params.term_crit = cvTermCriteria(CV_TERMCRIT_ITER, 1000, 0.01);
params.termCrit = TermCriteria(TermCriteria::MAX_ITER+TermCriteria::EPS, 1000, 0.01);
find_decision_boundary_SVM( params );
namedWindow( "classificationSVM1", WINDOW_AUTOSIZE );
imshow( "classificationSVM1", imgDst );
params.C = 10;
find_decision_boundary_SVM( params );
namedWindow( "classificationSVM2", WINDOW_AUTOSIZE );
imshow( "classificationSVM2", imgDst );
#endif
#if _DT_
find_decision_boundary_DT();
namedWindow( "DT", WINDOW_AUTOSIZE );
imshow( "DT", imgDst );
#endif
#if _BT_
find_decision_boundary_BT();
namedWindow( "BT", WINDOW_AUTOSIZE );
imshow( "BT", imgDst);
#endif
#if _GBT_
find_decision_boundary_GBT();
namedWindow( "GBT", WINDOW_AUTOSIZE );
imshow( "GBT", imgDst);
#endif
#if _RF_
find_decision_boundary_RF();
namedWindow( "RF", WINDOW_AUTOSIZE );
imshow( "RF", imgDst);
#endif
#if _ERT_
find_decision_boundary_ERT();
namedWindow( "ERT", WINDOW_AUTOSIZE );
imshow( "ERT", imgDst);
#endif
#if _ANN_
Mat layer_sizes1( 1, 3, CV_32SC1 );
layer_sizes1.at<int>(0) = 2;
layer_sizes1.at<int>(1) = 5;
layer_sizes1.at<int>(2) = classColors.size();
layer_sizes1.at<int>(2) = (int)classColors.size();
find_decision_boundary_ANN( layer_sizes1 );
namedWindow( "ANN", WINDOW_AUTOSIZE );
imshow( "ANN", imgDst );
#endif
#if _EM_
find_decision_boundary_EM();
namedWindow( "EM", WINDOW_AUTOSIZE );
imshow( "EM", imgDst );
#endif
}
samples/cpp/train_HOG.cpp
+26 -52
View File
@@ -8,9 +8,10 @@
#include <time.h>
using namespace cv;
using namespace cv::ml;
using namespace std;
void get_svm_detector(const SVM& svm, vector< float > & hog_detector );
void get_svm_detector(const Ptr<SVM>& svm, vector< float > & hog_detector );
void convert_to_ml(const std::vector< cv::Mat > & train_samples, cv::Mat& trainData );
void load_images( const string & prefix, const string & filename, vector< Mat > & img_lst );
void sample_neg( const vector< Mat > & full_neg_lst, vector< Mat > & neg_lst, const Size & size );
@@ -20,49 +21,24 @@ void train_svm( const vector< Mat > & gradient_lst, const vector< int > & labels
void draw_locations( Mat & img, const vector< Rect > & locations, const Scalar & color );
void test_it( const Size & size );
void get_svm_detector(const SVM& svm, vector< float > & hog_detector )
void get_svm_detector(const Ptr<SVM>& svm, vector< float > & hog_detector )
{
// get the number of variables
const int var_all = svm.get_var_count();
// get the number of support vectors
const int sv_total = svm.get_support_vector_count();
// get the decision function
const CvSVMDecisionFunc* decision_func = svm.get_decision_function();
// get the support vectors
const float** sv = new const float*[ sv_total ];
for( int i = 0 ; i < sv_total ; ++i )
sv[ i ] = svm.get_support_vector(i);
Mat sv = svm->getSupportVectors();
const int sv_total = sv.rows;
// get the decision function
Mat alpha, svidx;
double rho = svm->getDecisionFunction(0, alpha, svidx);
CV_Assert( var_all > 0 &&
sv_total > 0 &&
decision_func != 0 &&
decision_func->alpha != 0 &&
decision_func->sv_count == sv_total );
CV_Assert( alpha.total() == 1 && svidx.total() == 1 && sv_total == 1 );
CV_Assert( (alpha.type() == CV_64F && alpha.at<double>(0) == 1.) ||
(alpha.type() == CV_32F && alpha.at<float>(0) == 1.f) );
CV_Assert( sv.type() == CV_32F );
hog_detector.clear();
float svi = 0.f;
hog_detector.clear(); //clear stuff in vector.
hog_detector.reserve( var_all + 1 ); //reserve place for memory efficiency.
/**
* hog_detector^i = \sum_j support_vector_j^i * \alpha_j
* hog_detector^dim = -\rho
*/
for( int i = 0 ; i < var_all ; ++i )
{
svi = 0.f;
for( int j = 0 ; j < sv_total ; ++j )
{
if( decision_func->sv_index != NULL ) // sometime the sv_index isn't store on YML/XML.
svi += (float)( sv[decision_func->sv_index[j]][i] * decision_func->alpha[ j ] );
else
svi += (float)( sv[j][i] * decision_func->alpha[ j ] );
}
hog_detector.push_back( svi );
}
hog_detector.push_back( (float)-decision_func->rho );
delete[] sv;
hog_detector.resize(sv.cols + 1);
memcpy(&hog_detector[0], sv.data, sv.cols*sizeof(hog_detector[0]));
hog_detector[sv.cols] = (float)-rho;
}
@@ -263,7 +239,7 @@ Mat get_hogdescriptor_visu(const Mat& color_origImg, vector<float>& descriptorVa
int mx = drawX + cellSize/2;
int my = drawY + cellSize/2;
rectangle(visu, Point((int)(drawX*zoomFac), (int)(drawY*zoomFac)), Point((int)((drawX+cellSize)*zoomFac), (int)((drawY+cellSize)*zoomFac)), CV_RGB(100,100,100), 1);
rectangle(visu, Point((int)(drawX*zoomFac), (int)(drawY*zoomFac)), Point((int)((drawX+cellSize)*zoomFac), (int)((drawY+cellSize)*zoomFac)), Scalar(100,100,100), 1);
// draw in each cell all 9 gradient strengths
for (int bin=0; bin<gradientBinSize; bin++)
@@ -288,7 +264,7 @@ Mat get_hogdescriptor_visu(const Mat& color_origImg, vector<float>& descriptorVa
float y2 = my + dirVecY * currentGradStrength * maxVecLen * scale;
// draw gradient visualization
line(visu, Point((int)(x1*zoomFac),(int)(y1*zoomFac)), Point((int)(x2*zoomFac),(int)(y2*zoomFac)), CV_RGB(0,255,0), 1);
line(visu, Point((int)(x1*zoomFac),(int)(y1*zoomFac)), Point((int)(x2*zoomFac),(int)(y2*zoomFac)), Scalar(0,255,0), 1);
} // for (all bins)
@@ -337,28 +313,26 @@ void compute_hog( const vector< Mat > & img_lst, vector< Mat > & gradient_lst, c
void train_svm( const vector< Mat > & gradient_lst, const vector< int > & labels )
{
SVM svm;
/* Default values to train SVM */
SVMParams params;
SVM::Params params;
params.coef0 = 0.0;
params.degree = 3;
params.term_crit.epsilon = 1e-3;
params.termCrit.epsilon = 1e-3;
params.gamma = 0;
params.kernel_type = SVM::LINEAR;
params.kernelType = SVM::LINEAR;
params.nu = 0.5;
params.p = 0.1; // for EPSILON_SVR, epsilon in loss function?
params.C = 0.01; // From paper, soft classifier
params.svm_type = SVM::EPS_SVR; // C_SVC; // EPSILON_SVR; // may be also NU_SVR; // do regression task
params.svmType = SVM::EPS_SVR; // C_SVC; // EPSILON_SVR; // may be also NU_SVR; // do regression task
Mat train_data;
convert_to_ml( gradient_lst, train_data );
clog << "Start training...";
svm.train( train_data, Mat( labels ), Mat(), Mat(), params );
Ptr<SVM> svm = StatModel::train<SVM>(train_data, ROW_SAMPLE, Mat(labels), params);
clog << "...[done]" << endl;
svm.save( "my_people_detector.yml" );
svm->save( "my_people_detector.yml" );
}
void draw_locations( Mat & img, const vector< Rect > & locations, const Scalar & color )
@@ -380,7 +354,7 @@ void test_it( const Size & size )
Scalar reference( 0, 255, 0 );
Scalar trained( 0, 0, 255 );
Mat img, draw;
SVM svm;
Ptr<SVM> svm;
HOGDescriptor hog;
HOGDescriptor my_hog;
my_hog.winSize = size;
@@ -388,7 +362,7 @@ void test_it( const Size & size )
vector< Rect > locations;
// Load the trained SVM.
svm.load( "my_people_detector.yml" );
svm = StatModel::load<SVM>( "my_people_detector.yml" );
// Set the trained svm to my_hog
vector< float > hog_detector;
get_svm_detector( svm, hog_detector );
samples/cpp/tree_engine.cpp
+40 -88
View File
@@ -1,63 +1,35 @@
#include "opencv2/ml/ml.hpp"
#include "opencv2/core/core_c.h"
#include "opencv2/core/core.hpp"
#include "opencv2/core/utility.hpp"
#include <stdio.h>
#include <string>
#include <map>
using namespace cv;
using namespace cv::ml;
static void help()
{
printf(
"\nThis sample demonstrates how to use different decision trees and forests including boosting and random trees:\n"
"CvDTree dtree;\n"
"CvBoost boost;\n"
"CvRTrees rtrees;\n"
"CvERTrees ertrees;\n"
"CvGBTrees gbtrees;\n"
"Call:\n\t./tree_engine [-r <response_column>] [-c] <csv filename>\n"
"\nThis sample demonstrates how to use different decision trees and forests including boosting and random trees.\n"
"Usage:\n\t./tree_engine [-r <response_column>] [-ts type_spec] <csv filename>\n"
"where -r <response_column> specified the 0-based index of the response (0 by default)\n"
"-c specifies that the response is categorical (it's ordered by default) and\n"
"-ts specifies the var type spec in the form ord[n1,n2-n3,n4-n5,...]cat[m1-m2,m3,m4-m5,...]\n"
"<csv filename> is the name of training data file in comma-separated value format\n\n");
}
static int count_classes(CvMLData& data)
static void train_and_print_errs(Ptr<StatModel> model, const Ptr<TrainData>& data)
{
cv::Mat r = cv::cvarrToMat(data.get_responses());
std::map<int, int> rmap;
int i, n = (int)r.total();
for( i = 0; i < n; i++ )
bool ok = model->train(data);
if( !ok )
{
float val = r.at<float>(i);
int ival = cvRound(val);
if( ival != val )
return -1;
rmap[ival] = 1;
printf("Training failed\n");
}
return (int)rmap.size();
}
static void print_result(float train_err, float test_err, const CvMat* _var_imp)
{
printf( "train error %f\n", train_err );
printf( "test error %f\n\n", test_err );
if (_var_imp)
else
{
cv::Mat var_imp = cv::cvarrToMat(_var_imp), sorted_idx;
cv::sortIdx(var_imp, sorted_idx, CV_SORT_EVERY_ROW + CV_SORT_DESCENDING);
printf( "variable importance:\n" );
int i, n = (int)var_imp.total();
int type = var_imp.type();
CV_Assert(type == CV_32F || type == CV_64F);
for( i = 0; i < n; i++)
{
int k = sorted_idx.at<int>(i);
printf( "%d\t%f\n", k, type == CV_32F ? var_imp.at<float>(k) : var_imp.at<double>(k));
}
printf( "train error: %f\n", model->calcError(data, false, noArray()) );
printf( "test error: %f\n\n", model->calcError(data, true, noArray()) );
}
printf("\n");
}
int main(int argc, char** argv)
@@ -69,14 +41,14 @@ int main(int argc, char** argv)
}
const char* filename = 0;
int response_idx = 0;
bool categorical_response = false;
std::string typespec;
for(int i = 1; i < argc; i++)
{
if(strcmp(argv[i], "-r") == 0)
sscanf(argv[++i], "%d", &response_idx);
else if(strcmp(argv[i], "-c") == 0)
categorical_response = true;
else if(strcmp(argv[i], "-ts") == 0)
typespec = argv[++i];
else if(argv[i][0] != '-' )
filename = argv[i];
else
@@ -88,52 +60,32 @@ int main(int argc, char** argv)
}
printf("\nReading in %s...\n\n",filename);
CvDTree dtree;
CvBoost boost;
CvRTrees rtrees;
CvERTrees ertrees;
CvGBTrees gbtrees;
const double train_test_split_ratio = 0.5;
CvMLData data;
Ptr<TrainData> data = TrainData::loadFromCSV(filename, 0, response_idx, response_idx+1, typespec);
CvTrainTestSplit spl( 0.5f );
if ( data.read_csv( filename ) == 0)
if( data.empty() )
{
data.set_response_idx( response_idx );
if(categorical_response)
data.change_var_type( response_idx, CV_VAR_CATEGORICAL );
data.set_train_test_split( &spl );
printf("======DTREE=====\n");
dtree.train( &data, CvDTreeParams( 10, 2, 0, false, 16, 0, false, false, 0 ));
print_result( dtree.calc_error( &data, CV_TRAIN_ERROR), dtree.calc_error( &data, CV_TEST_ERROR ), dtree.get_var_importance() );
if( categorical_response && count_classes(data) == 2 )
{
printf("======BOOST=====\n");
boost.train( &data, CvBoostParams(CvBoost::DISCRETE, 100, 0.95, 2, false, 0));
print_result( boost.calc_error( &data, CV_TRAIN_ERROR ), boost.calc_error( &data, CV_TEST_ERROR ), 0 ); //doesn't compute importance
}
printf("======RTREES=====\n");
rtrees.train( &data, CvRTParams( 10, 2, 0, false, 16, 0, true, 0, 100, 0, CV_TERMCRIT_ITER ));
print_result( rtrees.calc_error( &data, CV_TRAIN_ERROR), rtrees.calc_error( &data, CV_TEST_ERROR ), rtrees.get_var_importance() );
printf("======ERTREES=====\n");
ertrees.train( &data, CvRTParams( 18, 2, 0, false, 16, 0, true, 0, 100, 0, CV_TERMCRIT_ITER ));
print_result( ertrees.calc_error( &data, CV_TRAIN_ERROR), ertrees.calc_error( &data, CV_TEST_ERROR ), ertrees.get_var_importance() );
printf("======GBTREES=====\n");
if (categorical_response)
gbtrees.train( &data, CvGBTreesParams(CvGBTrees::DEVIANCE_LOSS, 100, 0.1f, 0.8f, 5, false));
else
gbtrees.train( &data, CvGBTreesParams(CvGBTrees::SQUARED_LOSS, 100, 0.1f, 0.8f, 5, false));
print_result( gbtrees.calc_error( &data, CV_TRAIN_ERROR), gbtrees.calc_error( &data, CV_TEST_ERROR ), 0 ); //doesn't compute importance
printf("ERROR: File %s can not be read\n", filename);
return 0;
}
else
printf("File can not be read");
data->setTrainTestSplitRatio(train_test_split_ratio);
printf("======DTREE=====\n");
Ptr<DTrees> dtree = DTrees::create(DTrees::Params( 10, 2, 0, false, 16, 0, false, false, Mat() ));
train_and_print_errs(dtree, data);
if( (int)data->getClassLabels().total() <= 2 ) // regression or 2-class classification problem
{
printf("======BOOST=====\n");
Ptr<Boost> boost = Boost::create(Boost::Params(Boost::GENTLE, 100, 0.95, 2, false, Mat()));
train_and_print_errs(boost, data);
}
printf("======RTREES=====\n");
Ptr<RTrees> rtrees = RTrees::create(RTrees::Params(10, 2, 0, false, 16, Mat(), false, 0, TermCriteria(TermCriteria::MAX_ITER, 100, 0)));
train_and_print_errs(rtrees, data);
return 0;
}
samples/cpp/tutorial_code/ml/introduction_to_svm/introduction_to_svm.cpp
+18 -18
View File
@@ -4,29 +4,29 @@
#include <opencv2/ml/ml.hpp>
using namespace cv;
using namespace cv::ml;
int main()
int main(int, char**)
{
// Data for visual representation
int width = 512, height = 512;
Mat image = Mat::zeros(height, width, CV_8UC3);
// Set up training data
float labels[4] = {1.0, -1.0, -1.0, -1.0};
Mat labelsMat(4, 1, CV_32FC1, labels);
int labels[4] = {1, -1, -1, -1};
Mat labelsMat(4, 1, CV_32SC1, labels);
float trainingData[4][2] = { {501, 10}, {255, 10}, {501, 255}, {10, 501} };
Mat trainingDataMat(4, 2, CV_32FC1, trainingData);
// Set up SVM's parameters
CvSVMParams params;
params.svm_type = CvSVM::C_SVC;
params.kernel_type = CvSVM::LINEAR;
params.term_crit = cvTermCriteria(CV_TERMCRIT_ITER, 100, 1e-6);
SVM::Params params;
params.svmType = SVM::C_SVC;
params.kernelType = SVM::LINEAR;
params.termCrit = TermCriteria(TermCriteria::MAX_ITER, 100, 1e-6);
// Train the SVM
CvSVM SVM;
SVM.train(trainingDataMat, labelsMat, Mat(), Mat(), params);
Ptr<SVM> svm = StatModel::train<SVM>(trainingDataMat, ROW_SAMPLE, labelsMat, params);
Vec3b green(0,255,0), blue (255,0,0);
// Show the decision regions given by the SVM
@@ -34,30 +34,30 @@ int main()
for (int j = 0; j < image.cols; ++j)
{
Mat sampleMat = (Mat_<float>(1,2) << j,i);
float response = SVM.predict(sampleMat);
float response = svm->predict(sampleMat);
if (response == 1)
image.at<Vec3b>(i,j) = green;
else if (response == -1)
image.at<Vec3b>(i,j) = blue;
image.at<Vec3b>(i,j) = blue;
}
// Show the training data
int thickness = -1;
int lineType = 8;
circle( image, Point(501, 10), 5, Scalar( 0, 0, 0), thickness, lineType);
circle( image, Point(255, 10), 5, Scalar(255, 255, 255), thickness, lineType);
circle( image, Point(501, 255), 5, Scalar(255, 255, 255), thickness, lineType);
circle( image, Point( 10, 501), 5, Scalar(255, 255, 255), thickness, lineType);
circle( image, Point(501, 10), 5, Scalar( 0, 0, 0), thickness, lineType );
circle( image, Point(255, 10), 5, Scalar(255, 255, 255), thickness, lineType );
circle( image, Point(501, 255), 5, Scalar(255, 255, 255), thickness, lineType );
circle( image, Point( 10, 501), 5, Scalar(255, 255, 255), thickness, lineType );
// Show support vectors
thickness = 2;
lineType = 8;
int c = SVM.get_support_vector_count();
Mat sv = svm->getSupportVectors();
for (int i = 0; i < c; ++i)
for (int i = 0; i < sv.rows; ++i)
{
const float* v = SVM.get_support_vector(i);
const float* v = sv.ptr<float>(i);
circle( image, Point( (int) v[0], (int) v[1]), 6, Scalar(128, 128, 128), thickness, lineType);
}
samples/cpp/tutorial_code/ml/non_linear_svms/non_linear_svms.cpp
+11 -11
View File
@@ -8,6 +8,7 @@
#define FRAC_LINEAR_SEP 0.9f // Fraction of samples which compose the linear separable part
using namespace cv;
using namespace cv::ml;
using namespace std;
static void help()
@@ -30,7 +31,7 @@ int main()
//--------------------- 1. Set up training data randomly ---------------------------------------
Mat trainData(2*NTRAINING_SAMPLES, 2, CV_32FC1);
Mat labels (2*NTRAINING_SAMPLES, 1, CV_32FC1);
Mat labels (2*NTRAINING_SAMPLES, 1, CV_32SC1);
RNG rng(100); // Random value generation class
@@ -71,16 +72,15 @@ int main()
labels.rowRange(NTRAINING_SAMPLES, 2*NTRAINING_SAMPLES).setTo(2); // Class 2
//------------------------ 2. Set up the support vector machines parameters --------------------
CvSVMParams params;
params.svm_type = SVM::C_SVC;
SVM::Params params;
params.svmType = SVM::C_SVC;
params.C = 0.1;
params.kernel_type = SVM::LINEAR;
params.term_crit = TermCriteria(CV_TERMCRIT_ITER, (int)1e7, 1e-6);
params.kernelType = SVM::LINEAR;
params.termCrit = TermCriteria(TermCriteria::MAX_ITER, (int)1e7, 1e-6);
//------------------------ 3. Train the svm ----------------------------------------------------
cout << "Starting training process" << endl;
CvSVM svm;
svm.train(trainData, labels, Mat(), Mat(), params);
Ptr<SVM> svm = StatModel::train<SVM>(trainData, ROW_SAMPLE, labels, params);
cout << "Finished training process" << endl;
//------------------------ 4. Show the decision regions ----------------------------------------
@@ -89,7 +89,7 @@ int main()
for (int j = 0; j < I.cols; ++j)
{
Mat sampleMat = (Mat_<float>(1,2) << i, j);
float response = svm.predict(sampleMat);
float response = svm->predict(sampleMat);
if (response == 1) I.at<Vec3b>(j, i) = green;
else if (response == 2) I.at<Vec3b>(j, i) = blue;
@@ -117,11 +117,11 @@ int main()
//------------------------- 6. Show support vectors --------------------------------------------
thick = 2;
lineType = 8;
int x = svm.get_support_vector_count();
Mat sv = svm->getSupportVectors();
for (int i = 0; i < x; ++i)
for (int i = 0; i < sv.rows; ++i)
{
const float* v = svm.get_support_vector(i);
const float* v = sv.ptr<float>(i);
circle( I, Point( (int) v[0], (int) v[1]), 6, Scalar(128, 128, 128), thick, lineType);
}