private void Actualize()
{
// create a corresponding BayesClassifierModule object in MATLAB
var noOfCategories = _classificationCategories.Count;
var cmd = ClassifierUniqueId + " = BayesClassifierModule(" + noOfCategories + ");";
MatlabInterface.Execute(cmd);
// generate a Bayesian Classifier for each 'Pattern Classification Input' object.
foreach (var pci in InputNodes)
{
var minVal = 0;
var maxVal = 0;
if (pci.GetType() == typeof(Variable))
{
var v = (Variable)pci;
minVal = v.Value.MinimumAllowableValue;
maxVal = v.Value.MaximumAllowableValue;
}
else if (pci.GetType() == typeof(BayesClassifierModule))
{
var b = (BayesClassifierModule)pci;
minVal = 1;
maxVal = b.ClassificationCategories.Count;
}
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".newInputNode(" + minVal + ", " + maxVal + ");");
}
// lock BayesClassifierModule
cmd = ClassifierUniqueId + " = " + ClassifierUniqueId + ".lock();";
MatlabInterface.Execute(cmd);
}
/// <summary>
///
/// </summary>
/// <param name="inputVariable"></param>
/// <param name="inputVariableValue"></param>
/// <returns></returns>
public ClassCategory GetDecision(PatternClassificationInput inputVariable, int inputVariableValue)
{
// MatlabInterface.Execute request
MatlabInterface.Execute("inputClassifierDecision = " + ClassifierUniqueId + ".InputNodeClassifiers(" + inputVariable.ClassifierMatlabIndex + ").getDecision(" + inputVariableValue + ");");
// fetch value and return
var intResult = MatlabInterface.MatlabVariableToInteger("inputClassifierDecision", true);
var c = _classificationCategories[intResult - 1];
return(c);
}
/// <summary>
///
/// </summary>
/// <param name="inputVariable"></param>
/// <param name="inputVariableValue"></param>
/// <param name="c"></param>
/// <param name="plot"></param>
public void AssociateWithCategory(PatternClassificationInput inputVariable, int inputVariableValue,
ClassCategory c, bool plot)
{
// associate input value with cateogiry, for the specified variable
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".InputNodeClassifiers(" + inputVariable.ClassifierMatlabIndex + ").createAssociation(" + inputVariableValue + ", " + c.MatlabIndex + ");");
// plot
if (plot)
{
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".InputNodeClassifiers(" + inputVariable.ClassifierMatlabIndex + ").plotLikelihoods();");
}
}
/// <summary>
/// Associate current values with a particular category.
/// </summary>
public void AssociateWithCategory(int[] inputValues, ClassCategory c, bool plot)
{
MatlabInterface.ArrayToMatlabVector(inputValues, "evaluateTheseValues", true);
// MatlabInterface.Execute request
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".associateValuesWithCategory(evaluateTheseValues, " + c.MatlabIndex + ");");
// show us the result - comment out of not needed
if (plot)
{
MatlabInterface.Execute(ClassifierUniqueId + ".plotAllLikelihoods();");
}
}
private void CaptureReferenceImage(bool cropImage)
{
if (cropImage)
{
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".captureRefImgAndCrop();");
}
else
{
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".captureRefImg();");
}
}
/// <summary>
///
/// </summary>
/// <param name="inputVariable"></param>
/// <param name="priorProbabilities"></param>
public void SetCategoryPriories(PatternClassificationInput inputVariable, double[] priorProbabilities)
{
var matlabArrayName = "newPrioriesForInputNo_" + inputVariable.ClassifierMatlabIndex;
MatlabInterface.ArrayToMatlabVector(priorProbabilities, matlabArrayName, true);
if (priorProbabilities.Length != _classificationCategories.Count)
{
throw new ApplicationException(
"Number of priories in array does not match number of classification categories!");
}
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".setCategoryPrioriesForInput(" + inputVariable.ClassifierMatlabIndex +
", " + matlabArrayName + ");");
// plot resultant priories of BC
MatlabInterface.Execute("plot(" + ClassifierUniqueId + ".InputNodeClassifiers(" + inputVariable.ClassifierMatlabIndex +
").priories);");
}
private ClassCategory GetDecisionFromMatlabObj()
{
// MatlabInterface.Execute request
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".getDecision(evaluateTheseValues);");
MatlabInterface.Execute("Decision = " + ClassifierUniqueId + ".Decision;");
// get result
//var categoryNumber = Cognition.MatlabVariableToInteger(ClassifierUniqueId + ".Decision", false)
var categoryNumber = MatlabInterface.MatlabVariableToInteger("Decision", false);
if (categoryNumber == 0)
{
return(null);
}
var cat = _classificationCategories[categoryNumber - 1];
return(cat);
}
/// <summary>
/// Associate current values with a particular category.
/// </summary>
public void AssociateWithCategory(ClassCategory c, bool plot)
{
if (c == null)
{
throw new ApplicationException("No Category Selected.");
}
// create array of values in MATLAB
CreateArrayOfPresentValues(true, "evaluateTheseValues");
// MatlabInterface.Execute request
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".associateValuesWithCategory(evaluateTheseValues, " + c.MatlabIndex + ");");
// show us the result - comment out of not needed
if (plot)
{
MatlabInterface.Execute(ClassifierUniqueId + ".plotAllLikelihoods();");
}
}
public void ProvideTrainingResponse(bool objectDetected)
{
if (!_waitingForTrainingResponse)
{
throw new ApplicationException("MATLAB SURFDetector object is not expecting a training response!");
}
var v = 0;
// get int representative
if (objectDetected)
{
v = 1;
}
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".trainClassifier(" + v + ");");
// reset flag
_waitingForTrainingResponse = false;
}
public void CaptureCompareTrain()
{
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".captureCompareTrain();");
}
/// <summary>
///
/// </summary>
private void ActualizeInMatlab()
{
MatlabInterface.Execute(ClassifierUniqueId + " = " + ClassifierUniqueId + ".SURFDetector(0);");
}