// akhir pengaturan camera 4
private void ProcessFrame(Image <Bgr, Byte> imageproc, int cameranumber)
{
var counter = 0;
var numface = 0;
var status = 0;
var dataShape = new ConvNetSharp.Volume.Shape(48, 48, 1, 1);
var data = new double[dataShape.TotalLength];
/* Check to see that there was a frame collected */
if (imageproc != null)
{
var grayframe = imageproc.Convert <Gray, byte>();
var faces = cascade.DetectMultiScale(grayframe, 1.1, 10, System.Drawing.Size.Empty); //the actual face detection happens here
foreach (var face in faces)
{
numface = numface + 1;
imageproc.Draw(face, new Bgr(System.Drawing.Color.Blue), 3); //the detected face(s) is highlighted here using a box that is drawn around it/them
result = imageproc.Copy(face).Convert <Gray, byte>().Resize(48, 48, Inter.Linear); //wajah yang akan di kenali ekspresinya
// convert image to mat
Mat matImage = new Mat();
matImage = result.Mat;
// create volume and fill volume with pixels
var emotion = BuilderInstance <double> .Volume.From(data, dataShape);
for (var i = 0; i < 48; i++)
{
for (var j = 0; j < 48; j++)
{
emotion.Set(i, j, 0, MatExtension.GetValue(matImage, i, j));
}
}
// feed the network with volume
var results = fernet.Forward(emotion);
var c = fernet.GetPrediction();
var frresults = frnet.Forward(emotion);
var d = frnet.GetPrediction();
// mengakses softmax layer
var softmaxLayer = fernet.Layers[fernet.Layers.Count - 1] as SoftmaxLayer;
var activation = softmaxLayer.OutputActivation;
var N = activation.Shape.GetDimension(3);
var C = activation.Shape.GetDimension(2);
// mengambil setiap confidence level dari setiap label
for (var k = 0; k < 7; k++)
{
net_output[k] = Math.Round(activation.Get(1, 1, (k + 1), 0) * 100);
}
// display prediction result
// emotion prediction
if (c[0] == 0)
{
emotionstring = emotion_labels[0];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Blue), 3);
}
else
if (c[0] == 1)
{
emotionstring = emotion_labels[1];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Red), 3);
}
else
if (c[0] == 2)
{
emotionstring = emotion_labels[2];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Yellow), 3);
}
else
if (c[0] == 3)
{
emotionstring = emotion_labels[3];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Green), 3);
}
else
if (c[0] == 4)
{
emotionstring = emotion_labels[4];
imageproc.Draw(face, new Bgr(System.Drawing.Color.HotPink), 3);
}
else
if (c[0] == 5)
{
emotionstring = emotion_labels[5];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Orange), 3);
}
else
if (c[0] == 6)
{
emotionstring = emotion_labels[6];
imageproc.Draw(face, new Bgr(System.Drawing.Color.Purple), 3);
}
// face recognition prediction
if (d[0] == 0)
{
CvInvoke.PutText(imageproc, fr_labels[0] + "-" + emotionstring, face.Location, FontFace.HersheyComplex, 0.75, new MCvScalar(0, 255, 0));
}
else
if (d[0] == 1)
{
CvInvoke.PutText(imageproc, fr_labels[1] + "-" + emotionstring, face.Location, FontFace.HersheyComplex, 0.75, new MCvScalar(0, 255, 0));
}
else
if (d[0] == 2)
{
CvInvoke.PutText(imageproc, fr_labels[2] + "-" + emotionstring, face.Location, FontFace.HersheyComplex, 0.75, new MCvScalar(0, 255, 0));
//updatearray(status, numface, counter);
// update array
while (status == 0)
{
if (val_net[numface - 1, counter] < 0)
{
val_net[numface - 1, counter] = net_output[0];
val_ang[numface - 1, counter] = net_output[1];
val_dis[numface - 1, counter] = net_output[2];
val_fea[numface - 1, counter] = net_output[3];
val_hap[numface - 1, counter] = net_output[4];
val_sad[numface - 1, counter] = net_output[5];
val_sur[numface - 1, counter] = net_output[6];
status = 1;
}
else
{
counter = counter + 1;
}
}
chart1(counter);
}
}
}
// display ke gui
if (cameranumber == 1)
{
Camera1.Source = BitmapSourceConvert.ToBitmapSource(imageproc);
}
else if (cameranumber == 2)
{
Camera2.Source = BitmapSourceConvert.ToBitmapSource(imageproc);
}
else if (cameranumber == 3)
{
Camera3.Source = BitmapSourceConvert.ToBitmapSource(imageproc);
}
else if (cameranumber == 4)
{
Camera4.Source = BitmapSourceConvert.ToBitmapSource(imageproc);
}
}
