ENH: added the possibility to initialize the centroids from a file

Modules/Learning/Unsupervised/include/otbSharkKMeansMachineLearningModel.h

@@ -128,6 +128,10 @@ public:
   itkGetMacro( Normalized, bool );
   itkSetMacro( Normalized, bool );
 
+  /** If true, normalized input data sample list */
+  itkGetMacro( CentroidFilename, std::string );
+  itkSetMacro( CentroidFilename, std::string );
+
 protected:
   /** Constructor */
   SharkKMeansMachineLearningModel();
@@ -152,6 +156,9 @@ private:
   SharkKMeansMachineLearningModel(const Self &) = delete;
   void operator=(const Self &) = delete;
 
+  bool InitializeCentroids();
+  
+
   // Parameters set by the user
   bool m_Normalized;
   unsigned int m_K;
@@ -162,6 +169,8 @@ private:
   /** Centroids results form kMeans */
   shark::Centroids m_Centroids;
 
+  /** Input centroid filename */
+  std::string m_CentroidFilename;
 
   /** shark Model could be SoftClusteringModel or HardClusteringModel */
   boost::shared_ptr<ClusteringModelType> m_ClusteringModel;

Modules/Learning/Unsupervised/include/otbSharkKMeansMachineLearningModel.hxx

@@ -40,6 +40,7 @@
 #include "shark/Models/Clustering/HardClusteringModel.h"
 #include "shark/Models/Clustering/SoftClusteringModel.h"
 #include "shark/Algorithms/Trainers/NormalizeComponentsUnitVariance.h"
+#include <shark/Data/Csv.h> //load the csv file
 
 #if defined(__GNUC__) || defined(__clang__)
 #pragma GCC diagnostic pop
@@ -66,6 +67,18 @@ SharkKMeansMachineLearningModel<TInputValue, TOutputValue>
 {
 }
 
+template<class TInputValue, class TOutputValue>
+bool 
+SharkKMeansMachineLearningModel<TInputValue, TOutputValue>
+::InitializeCentroids()
+{
+  shark::Data<shark::RealVector> data;
+  shark::importCSV(data, m_CentroidFilename, ' ');
+  m_Centroids.setCentroids(data);
+  std::cout <<m_Centroids.centroids() << std::endl;
+  return 1;
+}
+
 /** Train the machine learning model */
 template<class TInputValue, class TOutputValue>
 void
@@ -77,6 +90,9 @@ SharkKMeansMachineLearningModel<TInputValue, TOutputValue>
   otb::Shark::ListSampleToSharkVector( this->GetInputListSample(), vector_data );
   shark::Data<shark::RealVector> data = shark::createDataFromRange( vector_data );
 
+  if (!m_CentroidFilename.empty())
+    InitializeCentroids();
+  
   // Normalized input value if necessary
   if( m_Normalized )
     data = NormalizeData( data );