amplement weidhting according to L. Bourdev and J. Brandt paper "Robust Object Detection Via Soft Cascade"

apps/sft/include/sft/octave.hpp

@@ -158,7 +158,7 @@ protected:
     float predict( const Mat& _sample, const cv::Range range) const;
 private:
     void traverse(const CvBoostTree* tree, cv::FileStorage& fs, const float* th = 0) const;
-
+    virtual void initial_weights(double (&p)[2]);
     cv::Rect boundingBox;
 
     int npositives;

apps/sft/octave.cpp

@@ -295,6 +295,7 @@ void sft::Octave::generateNegatives(const Dataset& dataset)
     dprintf("Processing negatives finished:\n\trequested %d negatives, viewed %d samples.\n", nnegatives, total);
 }
 
+
 template <typename T> int sgn(T val) {
     return (T(0) < val) - (val < T(0));
 }
@@ -378,6 +379,13 @@ void sft::Octave::write( cv::FileStorage &fso, const Mat& thresholds) const
         << "}";
 }
 
+void sft::Octave::initial_weights(double (&p)[2])
+{
+    double n = data->sample_count;
+    p[0] =  n / (double)(nnegatives) ;
+    p[1] =  n / (double)(npositives);
+}
+
 bool sft::Octave::train(const Dataset& dataset, const FeaturePool& pool, int weaks, int treeDepth)
 {
     CV_Assert(treeDepth == 2);

modules/ml/include/opencv2/ml/ml.hpp

@@ -1251,6 +1251,8 @@ protected:
     virtual void write_params( CvFileStorage* fs ) const;
     virtual void read_params( CvFileStorage* fs, CvFileNode* node );
 
+    virtual void initial_weights(double (&p)[2]);
+
     CvDTreeTrainData* data;
     CvBoostParams params;
     CvSeq* weak;

modules/ml/src/boost.cpp

@@ -1116,6 +1116,12 @@ bool CvBoost::train( CvMLData* _data,
     return result;
 }
 
+void CvBoost::initial_weights(double (&p)[2])
+{
+    p[0] = 1.;
+    p[1] = 1.;
+}
+
 void
 CvBoost::update_weights( CvBoostTree* tree )
 {
@@ -1159,8 +1165,9 @@ CvBoost::update_weights( CvBoostTree* tree )
         // in case of logitboost and gentle adaboost each weak tree is a regression tree,
         // so we need to convert class labels to floating-point values
 
-        double w0 = 1./n;
+        double w0 = 1./ n;
-        double p[2] = { 1, 1 };
+        double p[2] = { 1., 1. };
+        initial_weights(p);
 
         cvReleaseMat( &orig_response );
         cvReleaseMat( &sum_response );