diff --git a/include/DimensionalityReductionModelFactory.txx b/include/DimensionalityReductionModelFactory.txx
index 0c05937d39c7ba448c1f43ec360d10f62fcaea80..585af208aaef19604247d1e48369e65729d3a704 100644
--- a/include/DimensionalityReductionModelFactory.txx
+++ b/include/DimensionalityReductionModelFactory.txx
@@ -21,6 +21,7 @@
#include "DimensionalityReductionModelFactory.h"
#include "otbConfigure.h"
+#include "SOMModelFactory.h"
#ifdef OTB_USE_SHARK
#include "AutoencoderModelFactory.h"
@@ -95,15 +96,12 @@ DimensionalityReductionModelFactory<TInputValue,TOutputValue>
{
itk::MutexLockHolder<itk::SimpleMutexLock> lockHolder(mutex);
+
-#ifdef OTB_USE_SHARK
+ RegisterFactory(SOMModelFactory<TInputValue,TOutputValue>::New());
- // using AutoencoderModelFactory = AutoencoderModelFactoryBase<TInputValue, TTargetValue, shark::Autoencoder< shark::TanhNeuron, shark::LinearNeuron>> {};
-
-
- //using TiedAutoencoderModelFactory = public AutoencoderModelFactoryBase<TInputValue, TTargetValue, shark::TiedAutoencoder< shark::TanhNeuron, shark::LinearNeuron>> {};
-
+#ifdef OTB_USE_SHARK
RegisterFactory(PCAModelFactory<TInputValue,TOutputValue>::New());
RegisterFactory(AutoencoderModelFactory<TInputValue,TOutputValue>::New());
RegisterFactory(TiedAutoencoderModelFactory<TInputValue,TOutputValue>::New());
@@ -136,7 +134,14 @@ DimensionalityReductionModelFactory<TInputValue,TOutputValue>
for (itFac = factories.begin(); itFac != factories.end() ; ++itFac)
{
-
+ // SOM
+ SOMModelFactory<TInputValue,TOutputValue> *somFactory =
+ dynamic_cast<SOMModelFactory<TInputValue,TOutputValue> *>(*itFac);
+ if (somFactory)
+ {
+ itk::ObjectFactoryBase::UnRegisterFactory(somFactory);
+ continue;
+ }
#ifdef OTB_USE_SHARK
// Autoencoder
@@ -155,7 +160,8 @@ DimensionalityReductionModelFactory<TInputValue,TOutputValue>
itk::ObjectFactoryBase::UnRegisterFactory(taeFactory);
continue;
}
-
+
+ // PCA
PCAModelFactory<TInputValue,TOutputValue> *pcaFactory =
dynamic_cast<PCAModelFactory<TInputValue,TOutputValue> *>(*itFac);
if (pcaFactory)
diff --git a/include/SOMModel.h b/include/SOMModel.h
index 1a469e44369509cf81ccfe97876839222306897e..b7adddd663494f7d01c00a66062931ed5c7dbaac 100644
--- a/include/SOMModel.h
+++ b/include/SOMModel.h
@@ -37,7 +37,7 @@ public:
typedef typename Superclass::ConfidenceSampleType ConfidenceSampleType;
typedef typename Superclass::ConfidenceListSampleType ConfidenceListSampleType;
- typedef SOMMap<itk::VariableLengthVector<TInputValue>,itk::Statistics::EuclideanDistanceMetric<itk::VariableLengthVector<TInputValue>>, 2> MapType;
+ typedef SOMMap<itk::VariableLengthVector<TInputValue>,itk::Statistics::EuclideanDistanceMetric<itk::VariableLengthVector<TInputValue>>, 3> MapType;
typedef typename MapType::SizeType SizeType;
typedef otb::SOM<InputListSampleType, MapType> EstimatorType;
diff --git a/include/SOMModel.txx b/include/SOMModel.txx
index 78df8415c8470d3fb365030606bb1a60e45f0e8f..5c6be685bae44fbc4525c87f62f03166464720b5 100644
--- a/include/SOMModel.txx
+++ b/include/SOMModel.txx
@@ -14,6 +14,7 @@ namespace otb
template <class TInputValue>
SOMModel<TInputValue>::SOMModel()
{
+ this->m_IsRegressionSupported = true;
}
@@ -37,13 +38,6 @@ void SOMModel<TInputValue>::Train()
estimator->SetBetaEnd(m_BetaEnd);
estimator->SetMaxWeight(m_MaxWeight);
//AddProcess(estimator,"Learning");
- std::cout << "list = " << m_ListSample << std::endl;
- std::cout << "size = " << m_MapSize << std::endl;
- std::cout << "neigsize = " << m_NeighborhoodSizeInit << std::endl;
- std::cout << "n iter = " << m_NumberOfIterations << std::endl;
- std::cout << "bi = " << m_BetaInit << std::endl;
- std::cout << "be = " << m_BetaEnd << std::endl;
- std::cout << "mw = " << m_MaxWeight << std::endl;
estimator->Update();
@@ -54,6 +48,14 @@ void SOMModel<TInputValue>::Train()
template <class TInputValue>
bool SOMModel<TInputValue>::CanReadFile(const std::string & filename)
{
+ try
+ {
+ this->Load(filename);
+ }
+ catch(...)
+ {
+ return false;
+ }
return true;
}
@@ -86,7 +88,7 @@ void SOMModel<TInputValue>::Load(const std::string & filename, const std::string
auto reader = otb::ImageFileReader<MapType>::New();
reader->SetFileName(filename);
reader->Update();
- std::cout << reader->GetOutput()->GetImageKeywordlist().GetMetadataByKey("MachineLearningModelType") << '\n';
+ //std::cout << reader->GetOutput()->GetImageKeywordlist().GetMetadataByKey("MachineLearningModelType") << '\n';
m_SOMMap = reader->GetOutput();
}
@@ -94,7 +96,19 @@ void SOMModel<TInputValue>::Load(const std::string & filename, const std::string
template <class TInputValue>
typename SOMModel<TInputValue>::TargetSampleType
SOMModel<TInputValue>::DoPredict(const InputSampleType & value, ConfidenceValueType *quality) const
-{
+{
+ unsigned int dimension =MapType::ImageDimension;
+ TargetSampleType target;
+ target.SetSize(dimension);
+
+ auto winner =m_SOMMap->GetWinner(value);
+
+
+ for (int i=0; i< dimension ;i++) {
+ target[i] = winner.GetElement(i);
+ }
+
+ return target;
}
} // namespace otb
diff --git a/include/SOMModelFactory.h b/include/SOMModelFactory.h
index 56c0b88c6c3d85414d8f0d95b93f64047d2fbd06..0a3a9dd7a33e8366aa70d37d75a3af568c9a22c5 100644
--- a/include/SOMModelFactory.h
+++ b/include/SOMModelFactory.h
@@ -1,5 +1,5 @@
-#ifndef PCAModelFactory_h
-#define PCAModelFactory_h
+#ifndef SOMModelFactory_h
+#define SOMModelFactory_h
#include "itkObjectFactoryBase.h"
@@ -9,11 +9,11 @@ namespace otb
{
template <class TInputValue, class TTargetValue>
-class ITK_EXPORT PCAModelFactory : public itk::ObjectFactoryBase
+class ITK_EXPORT SOMModelFactory : public itk::ObjectFactoryBase
{
public:
/** Standard class typedefs. */
- typedef PCAModelFactory Self;
+ typedef SOMModelFactory Self;
typedef itk::ObjectFactoryBase Superclass;
typedef itk::SmartPointer<Self> Pointer;
typedef itk::SmartPointer<const Self> ConstPointer;
@@ -26,21 +26,21 @@ public:
itkFactorylessNewMacro(Self);
/** Run-time type information (and related methods). */
- itkTypeMacro(PCAModelFactory, itk::ObjectFactoryBase);
+ itkTypeMacro(SOMModelFactory, itk::ObjectFactoryBase);
/** Register one factory of this type */
static void RegisterOneFactory(void)
{
- Pointer PCAFactory = PCAModelFactory::New();
- itk::ObjectFactoryBase::RegisterFactory(PCAFactory);
+ Pointer SOMFactory = SOMModelFactory::New();
+ itk::ObjectFactoryBase::RegisterFactory(SOMFactory);
}
protected:
- PCAModelFactory();
- ~PCAModelFactory() ITK_OVERRIDE;
+ SOMModelFactory();
+ ~SOMModelFactory() ITK_OVERRIDE;
private:
- PCAModelFactory(const Self &); //purposely not implemented
+ SOMModelFactory(const Self &); //purposely not implemented
void operator =(const Self&); //purposely not implemented
};
@@ -51,7 +51,7 @@ private:
#ifndef OTB_MANUAL_INSTANTIATION
-#include "PCAModelFactory.txx"
+#include "SOMModelFactory.txx"
#endif
#endif
diff --git a/include/SOMModelFactory.txx b/include/SOMModelFactory.txx
index bfaa4f6f624b12d8351defb1bc52ad9a4d37253e..0df39c43c49d3983060c03f01572a3b5d417e129 100644
--- a/include/SOMModelFactory.txx
+++ b/include/SOMModelFactory.txx
@@ -15,49 +15,49 @@
PURPOSE. See the above copyright notices for more information.
=========================================================================*/
-#ifndef PCAFactory_txx
-#define PCAFactory_txx
+#ifndef SOMFactory_txx
+#define SOMFactory_txx
-#include "PCAModelFactory.h"
+#include "SOMModelFactory.h"
#include "itkCreateObjectFunction.h"
-#include "PCAModel.h"
+#include "SOMModel.h"
//#include <shark/Algorithms/Trainers/PCA.h>
#include "itkVersion.h"
namespace otb
{
template <class TInputValue, class TOutputValue>
-PCAModelFactory<TInputValue,TOutputValue>::PCAModelFactory()
+SOMModelFactory<TInputValue,TOutputValue>::SOMModelFactory()
{
std::string classOverride = std::string("DimensionalityReductionModel");
- std::string subclass = std::string("PCAModel");
+ std::string subclass = std::string("SOMModel");
this->RegisterOverride(classOverride.c_str(),
subclass.c_str(),
- "Shark PCA ML Model",
+ "SOM DR Model",
1,
// itk::CreateObjectFunction<AutoencoderModel<TInputValue,TOutputValue> >::New());
- itk::CreateObjectFunction<PCAModel<TInputValue>>::New());
+ itk::CreateObjectFunction<SOMModel<TInputValue>>::New());
}
template <class TInputValue, class TOutputValue>
-PCAModelFactory<TInputValue,TOutputValue>::~PCAModelFactory()
+SOMModelFactory<TInputValue,TOutputValue>::~SOMModelFactory()
{
}
template <class TInputValue, class TOutputValue>
-const char* PCAModelFactory<TInputValue,TOutputValue>::GetITKSourceVersion(void) const
+const char* SOMModelFactory<TInputValue,TOutputValue>::GetITKSourceVersion(void) const
{
return ITK_SOURCE_VERSION;
}
template <class TInputValue, class TOutputValue>
-const char* PCAModelFactory<TInputValue,TOutputValue>::GetDescription() const
+const char* SOMModelFactory<TInputValue,TOutputValue>::GetDescription() const
{
- return "PCA model factory";
+ return "SOM model factory";
}
} // end namespace otb
diff --git a/include/cbLearningApplicationBaseDR.h b/include/cbLearningApplicationBaseDR.h
index 9ef37ef948487c3c2f729d9a0b9e973af8de4d95..5ff2e90442d4630d8aed0d348c8eec70060ba56b 100644
--- a/include/cbLearningApplicationBaseDR.h
+++ b/include/cbLearningApplicationBaseDR.h
@@ -87,7 +87,7 @@ public:
// Dimensionality reduction models
- typedef SOMMap<itk::VariableLengthVector<TInputValue>,itk::Statistics::EuclideanDistanceMetric<itk::VariableLengthVector<TInputValue>>, 2> MapType;
+ typedef SOMMap<itk::VariableLengthVector<TInputValue>,itk::Statistics::EuclideanDistanceMetric<itk::VariableLengthVector<TInputValue>>, 3> MapType;
typedef otb::SOM<ListSampleType, MapType> EstimatorType;
typedef otb::SOMModel<InputValueType> SOMModelType;
diff --git a/include/cbTrainSOM.txx b/include/cbTrainSOM.txx
index 9e2b620dc52349f5a2f16eec97501b44d5998580..4951b177250802a4f097ebd10f897698a5472c6b 100644
--- a/include/cbTrainSOM.txx
+++ b/include/cbTrainSOM.txx
@@ -20,7 +20,14 @@ cbLearningApplicationBaseDR<TInputValue,TOutputValue>
"This group of parameters allows setting SOM parameters. "
);
-
+ AddParameter(ParameterType_StringList , "model.som.s", "Size");
+ SetParameterDescription("model.som.s", "Size of the SOM map");
+ MandatoryOff("model.som.s");
+
+ AddParameter(ParameterType_StringList , "model.som.n", "Size Neighborhood");
+ SetParameterDescription("model.som.n", "Size of the initial neighborhood in the SOM map");
+ MandatoryOff("model.som.n");
+
AddParameter(ParameterType_Int, "model.som.sx", "SizeX");
SetParameterDescription("model.som.sx", "X size of the SOM map");
MandatoryOff("model.som.sx");
@@ -52,7 +59,7 @@ cbLearningApplicationBaseDR<TInputValue,TOutputValue>
AddParameter(ParameterType_Float, "model.som.iv", "InitialValue");
SetParameterDescription("model.som.iv", "Maximum initial neuron weight");
MandatoryOff("model.som.iv");
-
+ ;
SetDefaultParameterInt("model.som.sx", 32);
SetDefaultParameterInt("model.som.sy", 32);
SetDefaultParameterInt("model.som.nx", 10);
@@ -70,20 +77,28 @@ void cbLearningApplicationBaseDR<TInputValue,TOutputValue>
::TrainSOM(typename ListSampleType::Pointer trainingListSample,std::string modelPath)
{
+ //std::cout << std::stoi(s[0]) << std::endl;
typename SOMModelType::Pointer dimredTrainer = SOMModelType::New();
dimredTrainer->SetNumberOfIterations(GetParameterInt("model.som.ni"));
dimredTrainer->SetBetaInit(GetParameterFloat("model.som.bi"));
dimredTrainer->SetBetaEnd(GetParameterFloat("model.som.bf"));
dimredTrainer->SetMaxWeight(GetParameterFloat("model.som.iv"));
+ std::cout << "0" << std::endl;
typename EstimatorType::SizeType size;
- size[0]=GetParameterInt("model.som.sx");
- size[1]=GetParameterInt("model.som.sy");
+ std::vector<std::basic_string<char>> s= GetParameterStringList("model.som.s");
+ for (int i=0; i<3; i++){ // This will be templated later (the 3)
+ size[i]=std::stoi(s[i]);
+ }
+
dimredTrainer->SetMapSize(size);
+ std::cout << "1" << std::endl;
typename EstimatorType::SizeType radius;
- radius[0] = GetParameterInt("model.som.nx");
- radius[1] = GetParameterInt("model.som.ny");
+ std::vector<std::basic_string<char>> n= GetParameterStringList("model.som.n");
+ for (int i=0; i<3; i++){ // This will be templated later (the 3)
+ radius[i]=std::stoi(n[i]);
+ }
+ std::cout << "2" << std::endl;
dimredTrainer->SetNeighborhoodSizeInit(radius);
- std::cout << trainingListSample << std::endl;
dimredTrainer->SetListSample(trainingListSample);
dimredTrainer->Train();
dimredTrainer->Save(modelPath);