using Microsoft.ML;

using Microsoft.ML.Data;

using System;

using System.IO;

using System.Linq;

namespace DeepNeuralNetwork

{

class Program

{

static void Main(string[] args)

{

var imagesFolder = Path.Combine(Environment.CurrentDirectory, "..", "..", "..", "images");

var files = Directory.GetFiles(imagesFolder, "*", SearchOption.AllDirectories);

var images = files.Select(file => new ImageData

{

ImagePath = file,

Label = Directory.GetParent(file).Name

});

var context = new MLContext();

var imageData = context.Data.LoadFromEnumerable(images);

var imageDataShuffled = context.Data.ShuffleRows(imageData);

var testTrainData = context.Data.TrainTestSplit(imageDataShuffled, testFraction: 0.2);

var validationData = context.Transforms.Conversion.MapValueToKey("LabelKey", "Label", keyOrdinality: Microsoft.ML.Transforms.ValueToKeyMappingEstimator.KeyOrdinality.ByValue)

.Fit(testTrainData.TestSet)

.Transform(testTrainData.TestSet);

var pipeline = context.Transforms.Conversion.MapValueToKey("LabelKey", "Label", keyOrdinality: Microsoft.ML.Transforms.ValueToKeyMappingEstimator.KeyOrdinality.ByValue)

.Append(context.Model.ImageClassification(

"ImagePath",

"LabelKey",

arch: Microsoft.ML.Transforms.ImageClassificationEstimator.Architecture.ResnetV2101,

epoch: 100,

batchSize: 10,

metricsCallback: Console.WriteLine,

validationSet: validationData));

var model = pipeline.Fit(testTrainData.TrainSet);

var predicions = model.Transform(testTrainData.TestSet);

var metrics = context.MulticlassClassification.Evaluate(predicions, labelColumnName: "LabelKey", predictedLabelColumnName: "PredictedLabel");

Console.WriteLine(Environment.NewLine);

Console.WriteLine($"Log loss - {metrics.LogLoss}");

var predictionEngine = context.Model.CreatePredictionEngine<ImageData, ImagePrediction>(model);

var testImagesFolder = Path.Combine(Environment.CurrentDirectory, "..", "..", "..", "test");

var testFiles = Directory.GetFiles(testImagesFolder, "*", SearchOption.AllDirectories);

var testImages = testFiles.Select(file => new ImageData

{

ImagePath = file

});

VBuffer<ReadOnlyMemory<char>> keys = default;

predictionEngine.OutputSchema["LabelKey"].GetKeyValues(ref keys);

var originalLabels = keys.DenseValues().ToArray();

Console.WriteLine(Environment.NewLine);

foreach (var image in testImages)

{

var prediction = predictionEngine.Predict(image);

var labelIndex = prediction.PredictedLabel;

Console.WriteLine($"Image : {Path.GetFileName(image.ImagePath)}, Score : {prediction.Score.Max()}, Predicted Label : {originalLabels[labelIndex]}");

}

context.Model.Save(model, imageData.Schema, "./dnn_model.zip");

Console.ReadLine();

}

}

}