otbConfusionMatrixMeasurements.hxx

Go to the documentation of this file.

/*
 * Copyright (C) 2005-2022 Centre National d'Etudes Spatiales (CNES)
 *
 * This file is part of Orfeo Toolbox
 *
 *     https://www.orfeo-toolbox.org/
 *
 * Licensed under the Apache License, Version 2.0 (the "License");
 * you may not use this file except in compliance with the License.
 * You may obtain a copy of the License at
 *
 *     http://www.apache.org/licenses/LICENSE-2.0
 *
 * Unless required by applicable law or agreed to in writing, software
 * distributed under the License is distributed on an "AS IS" BASIS,
 * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
 * See the License for the specific language governing permissions and
 * limitations under the License.
 */
 
#ifndef otbConfusionMatrixMeasurements_hxx
#define otbConfusionMatrixMeasurements_hxx
 
#include "otbConfusionMatrixMeasurements.h"
 
namespace otb
{
template <class TConfusionMatrix, class TLabel>
ConfusionMatrixMeasurements<TConfusionMatrix, TLabel>::ConfusionMatrixMeasurements()
  : m_KappaIndex(0.0),
    m_OverallAccuracy(0.0),
    m_FalseNegativeValue(0.0),
    m_TrueNegativeValue(0.0),
    m_FalsePositiveValue(0.0),
    m_TruePositiveValue(0.0),
    m_NumberOfClasses(0),
    m_NumberOfSamples(0)
 
{
  m_ConfusionMatrix = ConfusionMatrixType(m_NumberOfClasses, m_NumberOfClasses);
  m_ConfusionMatrix.Fill(0);
}
 
 
template <class TConfusionMatrix, class TLabel>
void ConfusionMatrixMeasurements<TConfusionMatrix, TLabel>::Compute()
{
  m_NumberOfClasses = m_ConfusionMatrix.Rows();
 
  m_FalseNegativeValues = MeasurementType(m_NumberOfClasses);
  m_TrueNegativeValues  = MeasurementType(m_NumberOfClasses);
  m_FalsePositiveValues = MeasurementType(m_NumberOfClasses);
  m_TruePositiveValues  = MeasurementType(m_NumberOfClasses);
  m_FalseNegativeValues.Fill(0);
  m_FalsePositiveValues.Fill(0);
  m_TruePositiveValues.Fill(0);
  m_TrueNegativeValues.Fill(0);
 
  const double epsilon    = 0.0000000001;
  double       luckyRate  = 0.;
  this->m_NumberOfSamples = 0;
  this->m_OverallAccuracy = 0.;
  for (unsigned int i = 0; i < m_NumberOfClasses; ++i)
  {
    double sum_along_row_i    = 0.;
    double sum_along_column_i = 0.;
    for (unsigned int j = 0; j < m_NumberOfClasses; ++j)
    {
      sum_along_row_i += m_ConfusionMatrix(i, j);
      sum_along_column_i += m_ConfusionMatrix(j, i);
      this->m_NumberOfSamples += m_ConfusionMatrix(i, j);
      // i != j
      if (i != j)
      {
        this->m_FalseNegativeValues[i] += m_ConfusionMatrix(i, j);
        this->m_FalsePositiveValues[i] += m_ConfusionMatrix(j, i);
      }
      // i == j
      else
      {
        this->m_OverallAccuracy += m_ConfusionMatrix(i, i);
        this->m_TruePositiveValues[i] = m_ConfusionMatrix(i, i);
      }
    }
    luckyRate += sum_along_row_i * sum_along_column_i;
  }
 
  this->m_TrueNegativeValues.Fill(m_NumberOfSamples);
  this->m_TrueNegativeValues -= this->m_FalseNegativeValues + this->m_FalsePositiveValues + this->m_TruePositiveValues;
 
  if (m_NumberOfClasses == 2)
  {
    this->m_TruePositiveValue  = m_ConfusionMatrix(0, 0);
    this->m_FalseNegativeValue = m_ConfusionMatrix(0, 1);
    this->m_FalsePositiveValue = m_ConfusionMatrix(1, 0);
    this->m_TrueNegativeValue  = m_ConfusionMatrix(1, 1);
    /*std::cout << "TP= " << this->m_TruePositiveValue << std::endl;
    std::cout << "FN= " << this->m_FalseNegativeValue << std::endl;
    std::cout << "FP= " << this->m_FalsePositiveValue << std::endl;
    std::cout << "TN= " << this->m_TrueNegativeValue << std::endl; */
  }
  else
  {
    /*std::cout << "TP= " << this->m_TruePositiveValues << std::endl;
    std::cout << "FN= " << this->m_FalseNegativeValues << std::endl;
    std::cout << "FP= " << this->m_FalsePositiveValues << std::endl;
    std::cout << "TN= " << this->m_TrueNegativeValues << std::endl; */
  }
 
  m_Precisions = MeasurementType(m_NumberOfClasses);
  m_Recalls    = MeasurementType(m_NumberOfClasses);
  m_FScores    = MeasurementType(m_NumberOfClasses);
  m_Precisions.Fill(0.);
  m_Recalls.Fill(0.);
  m_FScores.Fill(0.);
 
  if (m_NumberOfSamples > 0)
  {
    this->m_OverallAccuracy /= static_cast<double>(m_NumberOfSamples);
    luckyRate /= std::pow(m_NumberOfSamples, 2.0);
 
    if (std::abs(1 - luckyRate) > epsilon)
    {
      m_KappaIndex = (m_OverallAccuracy - luckyRate) / (1 - luckyRate);
    }
    else
    {
      this->m_KappaIndex = 1.;
    }
  }
  else
  {
    this->m_OverallAccuracy = 0.;
    this->m_KappaIndex      = 0.;
    // no samples, no need to continue
    return;
  }
 
  for (unsigned int i = 0; i < m_NumberOfClasses; ++i)
  {
    if (std::abs(this->m_TruePositiveValues[i] + this->m_FalsePositiveValues[i]) > epsilon)
    {
      this->m_Precisions[i] = this->m_TruePositiveValues[i] / (this->m_TruePositiveValues[i] + this->m_FalsePositiveValues[i]);
    }
 
    if (std::abs(this->m_TruePositiveValues[i] + this->m_FalseNegativeValues[i]) > epsilon)
    {
      this->m_Recalls[i] = this->m_TruePositiveValues[i] / (this->m_TruePositiveValues[i] + this->m_FalseNegativeValues[i]);
    }
 
    if (std::abs(this->m_Recalls[i] + this->m_Precisions[i]) > epsilon)
    {
      this->m_FScores[i] = 2 * this->m_Recalls[i] * this->m_Precisions[i] / (this->m_Recalls[i] + this->m_Precisions[i]);
    }
  }
 
 
  if (m_NumberOfClasses == 2)
  {
    if (std::abs(this->m_TruePositiveValue + this->m_FalsePositiveValue) > epsilon)
    {
      this->m_Precision = this->m_TruePositiveValue / (this->m_TruePositiveValue + this->m_FalsePositiveValue);
    }
    if (std::abs(this->m_TruePositiveValue + this->m_FalseNegativeValue) > epsilon)
    {
      this->m_Recall = this->m_TruePositiveValue / (this->m_TruePositiveValue + this->m_FalseNegativeValue);
    }
    if (std::abs(this->m_Recall + this->m_Precision) > epsilon)
    {
      this->m_FScore = 2 * this->m_Recall * this->m_Precision / (this->m_Recall + this->m_Precision);
    }
  }
}
 
/*
template <class TConfusionMatrix, class TLabel>
void
ConfusionMatrixMeasurements<TConfusionMatrix, TLabel>
::PrintSelf(std::ostream& os, itk::Indent indent) const
{
  os << indent << "TODO";
}*/
}
 
#endif