private void addAllImagesToDBToolStripMenuItem_Click(object sender, EventArgs e)
{
humanList = new List <Human>();
if (FBD.ShowDialog() == DialogResult.OK)
{
DirectoryInfo myFolder = new DirectoryInfo(FBD.SelectedPath);
fileCount = Directory.GetFiles(FBD.SelectedPath).Count();
InitializeProgressBar();
foreach (string filename in Directory.GetFiles(FBD.SelectedPath))
{
bitmap = new Bitmap(filename);
hog = new HistogramsOfOrientedGradients();
hog.ProcessImage(bitmap);
human = new Human();
if (filename.Contains("image_human"))
{
human.IsHuman = 1;
}
else
{
human.IsHuman = -1;
}
line = AuxiliaryFunctions.ToOneLine(hog.Histograms);
byteArray = AuxiliaryFunctions.DoubleArrayToByte(line);
human.HOG = byteArray;
//XML file
humanList.Add(human);
index++;
if (index % (fileCount / 100) == 0)
{
progressValue++;
label1.Text = (progressValue + " %").ToString();
pb.Value = progressValue <= 100?progressValue:100;
Thread.Sleep(10);
}
}
if (!File.Exists("Database.xml"))
{
AuxiliaryFunctions.Serialize(humanList, "Database.xml");
}
else
{
humanList.AddRange(AuxiliaryFunctions.Deserialize <List <Human> >("Database.xml"));
AuxiliaryFunctions.Serialize(humanList, "Database.xml");
}
}
}
private void trainHumansToolStripMenuItem_Click_1(object sender, EventArgs e)
{
//XML file
humanList = new List <Human>();
humanList = AuxiliaryFunctions.Deserialize <List <Human> >("Database.xml");
trainArray = new double[humanList.Count][];
outputArray = new double[humanList.Count];
fileCount = humanList.Count;
InitializeProgressBar();
for (int i = 0; i < humanList.Count; i++)
{
trainArray[i] = AuxiliaryFunctions.ByteArrayToDouble(humanList[i].HOG);
outputArray[i] = humanList[i].IsHuman;
index++;
if (index % (fileCount / 100) == 0)
{
progressValue++;
label1.Text = (progressValue + " %").ToString();
pb.Value = progressValue <= 100 ? progressValue : 100;
Thread.Sleep(10);
}
}
//trainArray = humanList.Select(x => AuxiliaryFunctions.ByteArrayToDouble(x.HOG)).ToArray();
//outputArray = humanList.Select(x => (double)x.IsHuman).ToArray();
var teacher = new SequentialMinimalOptimization <Gaussian>()
{
UseComplexityHeuristic = true,
UseKernelEstimation = true
};
SupportVectorMachine <Gaussian> svm = teacher.Learn(trainArray, outputArray);
resultLine = svm.Weights;
AuxiliaryFunctions.WriteWeight(resultLine, "weight.txt");
AuxiliaryFunctions.MakeSerialization(svm, "SVM_G.xml");
}
