static List <InputDataImages> ReadCSVFile(string path)
{
using (var reader = new StreamReader(@path))
{
List <InputDataImages> listA = new List <InputDataImages>();
//List<string> listB = new List<string>();
reader.ReadLine();
while (!reader.EndOfStream)
{
var line = reader.ReadLine();
var values = line.Split(',');
InputDataImages inputClass = new InputDataImages();
inputClass.Label = values[0].ToString();
inputClass.LeftEyebrow = float.Parse(values[1], System.Globalization.CultureInfo.InvariantCulture);
inputClass.RightEyebrow = float.Parse(values[2], System.Globalization.CultureInfo.InvariantCulture);
inputClass.LeftLip = float.Parse(values[3], System.Globalization.CultureInfo.InvariantCulture);
inputClass.RightLip = float.Parse(values[4], System.Globalization.CultureInfo.InvariantCulture);
inputClass.LipWidth = float.Parse(values[5], System.Globalization.CultureInfo.InvariantCulture);
inputClass.LipHeight = float.Parse(values[6], System.Globalization.CultureInfo.InvariantCulture);
listA.Add(inputClass);
}
return(listA);
}
}
static void Main(string[] args)
{
var mlContext = new MLContext();
// Need to create a DataView (FOR IMAGES)
IDataView dataView = mlContext.Data.LoadFromTextFile <InputDataImages>("feature_vectors.csv",
hasHeader: true, separatorChar: ',');
// Defining the models pipeline
var FeatureVectorName = "Features";
var LabelColumnName = "Label";
var pipeline = mlContext.Transforms.
Conversion.MapValueToKey(inputColumnName: "Label", outputColumnName: LabelColumnName)
.Append(mlContext.Transforms.Concatenate(FeatureVectorName, "LeftEyebrow", "RightEyebrow", "LeftLip", "RightLip", "LipWidth", "LipHeight"))
.AppendCacheCheckpoint(mlContext).
Append(mlContext.MulticlassClassification.Trainers.SdcaMaximumEntropy(LabelColumnName, FeatureVectorName))
.Append(mlContext.Transforms.Conversion.MapKeyToValue("PredictedLabel"));
var model = pipeline.Fit(dataView);
using (var fileStream = new FileStream("model.zip", FileMode.Create, FileAccess.Write, FileShare.Write))
{ mlContext.Model.Save(model, dataView.Schema, fileStream); }
//Define Predictor
var predictor = mlContext.Model.CreatePredictionEngine <InputDataImages, SpeciesPrediction>(model);
// Create list to store paths for images
List <string> Paths = Directory.GetFiles("Neutral", "*.png").ToList();
//string[] NeutralImagesPaths = Directory.GetFiles("TestingSetNeutral", "*.png").ToArray();
Paths.AddRange(Directory.GetFiles("Fear", "*.png").ToList());
Paths.AddRange(Directory.GetFiles("Disgust", "*.png").ToList());
Paths.AddRange(Directory.GetFiles("Anger", "*.png").ToList());
// Create header for CSV file
string Header = "Label, Score1, Score2, Score3, Score4, Path \n";
string HeaderImages = "Label, LeftEyebrow, RightEyebrow, LeftLip, RightLip, LipWidth, LipHeight \n";
System.IO.File.WriteAllText(@"TestingResults.csv", Header);
System.IO.File.WriteAllText(@"TestingFeatureVectorValues.csv", HeaderImages);
// Load Image and compute feature values
//string str = "S116_006_02240709.png";
foreach (var str in Paths)
{
InputDataImages inputValues = GetFeaturesValuesFromImage(str);
//compute label from parent's directory name
//var prediction = predictor.Predict(new InputDataImages()
//{
// LeftEyebrow = inputValues.LeftEyebrow,
// RightEyebrow = inputValues.RightEyebrow,
// LeftLip = inputValues.LeftLip,
// RightLip = inputValues.RightLip,
// LipWidth = inputValues.LipWidth,
// LipHeight = inputValues.LipHeight
//});
//Console.Write($"*** Prediction: {prediction.Species} ***");
//Console.Write($"*** Scores: {string.Join(" ", prediction.Scores)} ***");
//Console.Write($"*** Path: {str} ***");
//Console.WriteLine();
}
List <InputDataImages> TestingFeatureVectorValues = ReadCSVFile("TestingFeatureVectorValues.csv");
foreach (var Input in TestingFeatureVectorValues)
{
var prediction = predictor.Predict(new InputDataImages()
{
LeftEyebrow = Input.LeftEyebrow,
RightEyebrow = Input.RightEyebrow,
LeftLip = Input.LeftLip,
RightLip = Input.RightLip,
LipWidth = Input.LipWidth,
LipHeight = Input.LipHeight
});
//Write resutls inside CSV file
using (System.IO.StreamWriter file = new System.IO.StreamWriter(@"TestingResults.csv", true))
{
file.WriteLine(prediction.Species.ToString() + "," + prediction.Scores[0].ToString()
+ "," + prediction.Scores[1].ToString() + "," + prediction.Scores[2].ToString() + "," +
prediction.Scores[3].ToString() + "," + Input.Label.ToString());
}
}
//InputDataImages inputValues = GetFeaturesValuesFromImage(str);
// SeinputFilePathst up Dlib Face Detector
//var prediction = predictor.Predict(new InputDataImages()
//{
// LeftEyebrow = inputValues.LeftEyebrow,
// RightEyebrow = inputValues.RightEyebrow,
// LeftLip = inputValues.LeftLip,
// RightLip = inputValues.RightLip,
// LipWidth = inputValues.LipWidth,
// LipHeight = inputValues.LipHeight
//});
// THIS IS JUST FOR GETTING THE MICROACCURACY
var TestDataView = mlContext.Data.LoadFromTextFile <InputDataImages>("TestingFeatureVectorValues.csv", hasHeader: true, separatorChar: ',');
var TestMetrics = mlContext.MulticlassClassification.Evaluate(model.Transform(dataView));
Console.WriteLine($"* Metrics for Multi-class Classification model - Test Data");
Console.WriteLine($"* MicroAccuracy: {TestMetrics.MicroAccuracy:0.###}");
Console.WriteLine($"* MacroAccuracy: {TestMetrics.MacroAccuracy:0.###}");
Console.WriteLine($"* LogLoss: {TestMetrics.LogLoss:#.###}");
Console.WriteLine($"* LogLossReduction: {TestMetrics.LogLossReduction:#.###}");
//Console.Write($"*** Prediction: {prediction.Species} ***");
//Console.Write($"*** Scores: {string.Join(" ", prediction.Scores)} ***");
}
private static InputDataImages GetFeaturesValuesFromImage(string str)
{
var returnClass = new InputDataImages();
using (var fd = Dlib.GetFrontalFaceDetector())
// ... and Dlib Shape DetectorS
using (var sp = ShapePredictor.Deserialize("shape_predictor_68_face_landmarks.dat"))
{
// load input image
var img = Dlib.LoadImage <RgbPixel>(str);
// find all faces i n the image
var faces = fd.Operator(img);
// for each face draw over the facial landmarks
// Create the CSV file and fill in the first line with the header
foreach (var face in faces)
{
// find the landmark points for this face
var shape = sp.Detect(img, face);
// draw the landmark points on the image
for (var i = 0; i < shape.Parts; i++)
{
var point = shape.GetPart((uint)i);
var rect = new Rectangle(point);
Dlib.DrawRectangle(img, rect, color: new RgbPixel(255, 255, 0), thickness: 4);
}
/////////////// WEEK 9 LAB ////////////////
double[] LeftEyebrowDistances = new double[4];
double[] RightEyebrowDistances = new double[4];
float LeftEyebrowSum = 0;
float RightEyebrowSum = 0;
//LIP VARIABLES
double[] LeftLipDistances = new double[4];
double[] RightLipDistances = new double[4];
float LeftLipSum = 0;
float RightLipSum = 0;
LeftEyebrowDistances[0] = (shape.GetPart(21) - shape.GetPart(39)).Length;
LeftEyebrowDistances[1] = (shape.GetPart(20) - shape.GetPart(39)).Length;
LeftEyebrowDistances[2] = (shape.GetPart(19) - shape.GetPart(39)).Length;
LeftEyebrowDistances[3] = (shape.GetPart(18) - shape.GetPart(39)).Length;
RightEyebrowDistances[0] = (shape.GetPart(22) - shape.GetPart(42)).Length;
RightEyebrowDistances[1] = (shape.GetPart(23) - shape.GetPart(42)).Length;
RightEyebrowDistances[2] = (shape.GetPart(24) - shape.GetPart(42)).Length;
RightEyebrowDistances[3] = (shape.GetPart(25) - shape.GetPart(42)).Length;
//LIP
LeftLipDistances[0] = (shape.GetPart(51) - shape.GetPart(33)).Length;
LeftLipDistances[1] = (shape.GetPart(50) - shape.GetPart(33)).Length;
LeftLipDistances[2] = (shape.GetPart(49) - shape.GetPart(33)).Length;
LeftLipDistances[3] = (shape.GetPart(48) - shape.GetPart(33)).Length;
RightLipDistances[0] = (shape.GetPart(51) - shape.GetPart(33)).Length;
RightLipDistances[1] = (shape.GetPart(52) - shape.GetPart(33)).Length;
RightLipDistances[2] = (shape.GetPart(53) - shape.GetPart(33)).Length;
RightLipDistances[3] = (shape.GetPart(54) - shape.GetPart(33)).Length;
for (int i = 0; i < 4; i++)
{
LeftEyebrowSum += (float)(LeftEyebrowDistances[i] / LeftEyebrowDistances[0]);
RightEyebrowSum += (float)(RightEyebrowDistances[i] / RightEyebrowDistances[0]);
}
LeftLipSum += (float)(LeftLipDistances[1] / LeftLipDistances[0]);
LeftLipSum += (float)(LeftLipDistances[2] / LeftLipDistances[0]);
LeftLipSum += (float)(LeftLipDistances[3] / LeftLipDistances[0]);
RightLipSum += (float)(RightLipDistances[1] / RightLipDistances[0]);
RightLipSum += (float)(RightLipDistances[2] / RightLipDistances[0]);
RightLipSum += (float)(RightLipDistances[3] / RightLipDistances[0]);
double LipWidth = (float)((shape.GetPart(48) - shape.GetPart(54)).Length / (shape.GetPart(33) - shape.GetPart(51)).Length);
double LipHeight = (float)((shape.GetPart(51) - shape.GetPart(57)).Length / (shape.GetPart(33) - shape.GetPart(51)).Length);
returnClass.LeftEyebrow = LeftEyebrowSum;
returnClass.RightEyebrow = RightLipSum;
returnClass.LeftLip = LeftLipSum;
returnClass.RightLip = RightLipSum;
returnClass.LipWidth = (float)LipWidth;
returnClass.LipHeight = (float)LipHeight;
// export the modified image
string filePath = "output" + ".jpg";
Dlib.SaveJpeg(img, filePath);
}
}
using (System.IO.StreamWriter file = new System.IO.StreamWriter(@"TestingFeatureVectorValues.csv", true))
{
DirectoryInfo dr = new DirectoryInfo(str);
//Console.WriteLine(dr.Parent.Name.ToString());
string ParentFolderName = dr.Parent.Name.ToString();
file.WriteLine(ParentFolderName + "," + returnClass.LeftEyebrow.ToString() + "," + returnClass.RightEyebrow.ToString()
+ "," + returnClass.LeftLip.ToString() + "," + returnClass.RightLip.ToString() + "," + returnClass.LipWidth.ToString()
+ "," + returnClass.LipHeight.ToString());
}
return(returnClass);
}
