private void BtnSkip_Click(object sender, RoutedEventArgs e)
{
Button button = sender as Button;
RecognitionModel model = button.Tag as RecognitionModel;
PV_Recognizer.Train(model, "Unknown");
icImages.ItemsSource = PeopleImage.Where(x => x.Processed == false);
}
private void BtnSave_Click(object sender, RoutedEventArgs e)
{
Button button = sender as Button;
RecognitionModel model = button.Tag as RecognitionModel;
TextBox textBox = (button.Parent as DockPanel).Children.OfType <TextBox>().FirstOrDefault();
string name = textBox.Text;
PV_Recognizer.Train(model, name);
icImages.ItemsSource = PeopleImage.Where(x => x.Processed == false);
}
public static void Train(RecognitionModel model, String name)
{
model.Processed = true;
if (dic_labels.Values.Contains(name))
{
model.LabelInt = dic_labels.FirstOrDefault(x => x.Value == name).Key;
}
else
{
int num = dic_labels.Count;
dic_labels.Add(num, name);
model.LabelInt = num;
}
}
public static List <RecognitionModel> Recognize(IInputArray image, String uri)
{
if (training_sets.Exists(x => x.FullImageUri == uri))
{
return(new List <RecognitionModel>());
}
List <Rectangle> faces = new List <Rectangle>();
//List<Rectangle> eyes = new List<Rectangle>();
long detectionTime;
Detect(image, "haarcascade_frontalface_default.xml", /*"haarcascade_eye.xml",*/ faces, /*eyes,*/ out detectionTime);
List <RecognitionModel> results = new List <RecognitionModel>();
foreach (Rectangle r in faces)
{
IImage face = new UMat((UMat)image, r);
Directory.CreateDirectory("training_set\\");
String path = "training_set\\" + Path.GetRandomFileName() + ".jpg";
new Image <Bgr, byte>(face.Bitmap).Resize(100, 100, Inter.Cubic).Save(path);
RecognitionModel rm = new RecognitionModel(0, path, uri, false);
if (Trained)
{
using (Image <Gray, Byte> f = new Image <Gray, byte>(face.Bitmap))
{
FaceRecognizer.PredictionResult predictionResults = Recognizer.Predict(f.Resize(100, 100, Inter.Cubic));
Console.WriteLine(predictionResults.Distance + " " + predictionResults.Label);
rm.Distance = predictionResults.Distance;
if (predictionResults.Distance < 6000 && dic_labels.ContainsKey(predictionResults.Label))
{
rm.LabelInt = predictionResults.Label;
}
else
{
rm.LabelInt = 0;
}
}
}
else
{
rm.LabelInt = 0;
}
results.Add(rm);
training_sets.Add(rm);
}
return(results);
}
public static List <RecognitionModel> Init()
{
training_sets = new List <RecognitionModel>();
dic_labels = new Dictionary <int, string>();
using (StreamReader reader = new StreamReader(new BufferedStream(new FileStream("labels.txt", FileMode.OpenOrCreate, FileAccess.Read), 512)))
{
while (!reader.EndOfStream)
{
string[] vals = reader.ReadLine().Split('|');
dic_labels.Add(int.Parse(vals[0]), vals[1]);
}
if (dic_labels.Count == 0)
{
dic_labels.Add(0, "Unknown");
}
}
using (StreamReader reader = new StreamReader(new BufferedStream(new FileStream("training_sets.txt", FileMode.OpenOrCreate, FileAccess.Read), 512)))
{
while (!reader.EndOfStream)
{
string[] vals = reader.ReadLine().Split('|');
RecognitionModel trainingSet = new RecognitionModel(int.Parse(vals[0]), vals[1], vals[2], int.Parse(vals[3]) == 0 ? false : true);
if (vals.Length > 4)
{
trainingSet.Distance = double.Parse(vals[4]);
}
training_sets.Add(trainingSet);
}
}
if (training_sets.Count > 0)
{
IEnumerable <RecognitionModel> arr = training_sets.Where(x => x.Processed);
Recognizer.Train <Gray, byte>(arr.Select(x => x.FaceImageCV).ToArray(), arr.Select(x => x.LabelInt).ToArray());
Trained = true;
}
Console.WriteLine(training_sets.Count);
return(training_sets);
}
