Delete function areClassesEmpty().

modules/ml/include/opencv2/ml.hpp

@@ -1507,7 +1507,7 @@ public:
 SVMSGD provides a fast and easy-to-use implementation of the SVM classifier using the Stochastic Gradient Descent approach,
 as presented in @cite bottou2010large.
 
-The classifier has 5 parameters. These are
+The classifier has following parameters:
 - model type,
 - margin type,
 - margin regularization (\f$\lambda\f$),
@@ -1567,11 +1567,8 @@ To use SVMSGD algorithm do as follows:
 // Create empty object
 cv::Ptr<SVMSGD> svmsgd = SVMSGD::create();
 
-// Set parameters
-svmsgd->setOptimalParameters();
-
 // Train the Stochastic Gradient Descent SVM
-SvmSgd->train(trainData);
+svmsgd->train(trainData);
 
 // Predict labels for the new samples
 svmsgd->predict(samples, responses);

modules/ml/src/svmsgd.cpp

@@ -99,8 +99,6 @@ public:
 private:
     void updateWeights(InputArray sample, bool isPositive, float stepSize, Mat &weights);
 
-    std::pair<bool,bool> areClassesEmpty(Mat responses);
-
     void writeParams( FileStorage &fs ) const;
 
     void readParams( const FileNode &fn );
@@ -138,26 +136,6 @@ Ptr<SVMSGD> SVMSGD::create()
     return makePtr<SVMSGDImpl>();
 }
 
-std::pair<bool,bool> SVMSGDImpl::areClassesEmpty(Mat responses)
-{
-    CV_Assert(responses.cols == 1 || responses.rows == 1);
-    std::pair<bool,bool> emptyInClasses(true, true);
-    int limitIndex = responses.rows;
-
-    for(int index = 0; index < limitIndex; index++)
-    {
-        if (isPositive(responses.at<float>(index)))
-            emptyInClasses.first = false;
-        else
-            emptyInClasses.second = false;
-
-        if (!emptyInClasses.first && ! emptyInClasses.second)
-            break;
-    }
-
-    return emptyInClasses;
-}
-
 void SVMSGDImpl::normalizeSamples(Mat &samples, Mat &average, float &multiplier)
 {
     int featuresCount = samples.cols;
@@ -248,16 +226,20 @@ bool SVMSGDImpl::train(const Ptr<TrainData>& data, int)
     int featureCount = trainSamples.cols;
     Mat trainResponses = data->getTrainResponses();        // (trainSamplesCount x 1) matrix
 
-    std::pair<bool,bool> areEmpty = areClassesEmpty(trainResponses);
+    CV_Assert(trainResponses.rows == trainSamples.rows);
 
-    if ( areEmpty.first && areEmpty.second )
+    if (trainResponses.empty())
     {
         return false;
     }
-    if ( areEmpty.first || areEmpty.second )
+
+    int positiveCount = countNonZero(trainResponses >= 0);
+    int negativeCount = countNonZero(trainResponses < 0);
+
+    if ( positiveCount <= 0 || negativeCount <= 0 )
     {
         weights_ = Mat::zeros(1, featureCount, CV_32F);
-        shift_ = areEmpty.first ? -1.f : 1.f;
+        shift_ = (positiveCount > 0) ? 1.f : -1.f;
         return true;
     }
 
@@ -340,7 +322,7 @@ float SVMSGDImpl::predict( InputArray _samples, OutputArray _results, int ) cons
     int nSamples = samples.rows;
     cv::Mat results;
 
-    CV_Assert( samples.cols == weights_.cols && samples.type() == CV_32F );
+    CV_Assert( samples.cols == weights_.cols && samples.type() == CV_32FC1);
 
     if( _results.needed() )
     {
@@ -498,17 +480,7 @@ void SVMSGDImpl::clear()
 SVMSGDImpl::SVMSGDImpl()
 {
     clear();
-
+    setOptimalParameters();
-    params.svmsgdType = -1;
-    params.marginType = -1;
-
-    // Parameters for learning
-    params.marginRegularization = 0;                              // regularization
-    params.initialStepSize = 0;                        // learning rate (ideally should be large at beginning and decay each iteration)
-    params.stepDecreasingPower = 0;
-
-    TermCriteria _termCrit(TermCriteria::COUNT + TermCriteria::EPS, 0, 0);
-    params.termCrit = _termCrit;
 }
 
 void SVMSGDImpl::setOptimalParameters(int svmsgdType, int marginType)

modules/ml/test/test_svmsgd.cpp

@@ -182,8 +182,6 @@ void CV_SVMSGDTrainTest::run( int /*start_from*/ )
 {
     cv::Ptr<SVMSGD> svmsgd = SVMSGD::create();
 
-    svmsgd->setOptimalParameters();
-
     svmsgd->train(data);
 
     Mat responses;

samples/cpp/train_svmsgd.cpp

@@ -46,7 +46,6 @@ void addPointRetrainAndRedraw(Data &data, int x, int y, int response);
 bool doTrain( const Mat samples, const Mat responses, Mat &weights, float &shift)
 {
     cv::Ptr<SVMSGD> svmsgd = SVMSGD::create();
-    svmsgd->setOptimalParameters();
 
     cv::Ptr<TrainData> trainData = TrainData::create(samples, cv::ml::ROW_SAMPLE, responses);
     svmsgd->train( trainData );