Check sure that we're not already below required leaf false alarm rate before continuing to get negative samples.

apps/traincascade/cascadeclassifier.cpp

@@ -198,7 +198,7 @@ bool CvCascadeClassifier::train( const string _cascadeDirName,
         cout << endl << "===== TRAINING " << i << "-stage =====" << endl;
         cout << "<BEGIN" << endl;
 
-        if ( !updateTrainingSet( tempLeafFARate ) )
+        if ( !updateTrainingSet( requiredLeafFARate, tempLeafFARate ) )
         {
             cout << "Train dataset for temp stage can not be filled. "
                 "Branch training terminated." << endl;
@@ -284,17 +284,17 @@ int CvCascadeClassifier::predict( int sampleIdx )
     return 1;
 }
 
-bool CvCascadeClassifier::updateTrainingSet( double& acceptanceRatio)
+bool CvCascadeClassifier::updateTrainingSet( double minimumAcceptanceRatio, double& acceptanceRatio)
 {
     int64 posConsumed = 0, negConsumed = 0;
     imgReader.restart();
-    int posCount = fillPassedSamples( 0, numPos, true, posConsumed );
+    int posCount = fillPassedSamples( 0, numPos, true, 0, posConsumed );
     if( !posCount )
         return false;
     cout << "POS count : consumed   " << posCount << " : " << (int)posConsumed << endl;
 
     int proNumNeg = cvRound( ( ((double)numNeg) * ((double)posCount) ) / numPos ); // apply only a fraction of negative samples. double is required since overflow is possible
-    int negCount = fillPassedSamples( posCount, proNumNeg, false, negConsumed );
+    int negCount = fillPassedSamples( posCount, proNumNeg, false, minimumAcceptanceRatio, negConsumed );
     if ( !negCount )
         return false;
 
@@ -304,7 +304,7 @@ bool CvCascadeClassifier::updateTrainingSet( double& acceptanceRatio)
     return true;
 }
 
-int CvCascadeClassifier::fillPassedSamples( int first, int count, bool isPositive, int64& consumed )
+int CvCascadeClassifier::fillPassedSamples( int first, int count, bool isPositive, double minimumAcceptanceRatio, int64& consumed )
 {
     int getcount = 0;
     Mat img(cascadeParams.winSize, CV_8UC1);
@@ -312,6 +312,9 @@ int CvCascadeClassifier::fillPassedSamples( int first, int count, bool isPositiv
     {
         for( ; ; )
         {
+            if( consumed != 0 && ((double)getcount+1)/(double)(int64)consumed <= minimumAcceptanceRatio )
+                return getcount;
+
             bool isGetImg = isPositive ? imgReader.getPos( img ) :
                                            imgReader.getNeg( img );
             if( !isGetImg )

apps/traincascade/cascadeclassifier.h

@@ -101,8 +101,8 @@ private:
     int predict( int sampleIdx );
     void save( const std::string cascadeDirName, bool baseFormat = false );
     bool load( const std::string cascadeDirName );
-    bool updateTrainingSet( double& acceptanceRatio );
+    bool updateTrainingSet( double minimumAcceptanceRatio, double& acceptanceRatio );
-    int fillPassedSamples( int first, int count, bool isPositive, int64& consumed );
+    int fillPassedSamples( int first, int count, bool isPositive, double requiredAcceptanceRatio, int64& consumed );
 
     void writeParams( cv::FileStorage &fs ) const;
     void writeStages( cv::FileStorage &fs, const cv::Mat& featureMap ) const;