/// <summary>
/// Trains the classifier to recognise a single letter based on an image from our 'sample-data' folder
/// </summary>
private static void TrainSampleCharacter(Classifier <char> classifier, char character)
{
WrappedBitmap wb = WrappedBitmap.FromFile($@"..\..\..\..\sample-data\{character}.png");
Console.Write(character);
classifier.Train(wb, character, TuningParams.TrainingRounds);
}
/// <summary>
/// Trains the model to recognise the specified input as the specified symbol
/// </summary>
public void Train(WrappedBitmap inputData, TSymbol representedValue, uint rounds)
{
var initialGeneration = Enumerable.Range(0, (int)rounds)
.Select(i => new Scorer(new TuningParams.Scoring())) // TODO nested constructors is smelly
.Select(s => new ScorerScorePair(s, s.Score(inputData)))
.AsParallel()
.OrderByDescending(s => s.Score);
// Having acquired the initial set of sampling pixels via sheer randomness, we'll work with
// the best we've got and try and massage them into something better.
var mutParams = new TuningParams.Mutation();
var initialBestCandidates = initialGeneration.Take(100); // TODO parameterise magic value (which was decided totally arbitrarily anyway)
var initialBestCandidatesWithMutatedVariants = from p in initialBestCandidates
let mutatedOffspring = p.Scorer.MutateMany(mutParams)
let mutatedScores = from m in mutatedOffspring
let mutatedScore = m.Score(inputData)
select new ScorerScorePair(m, mutatedScore)
select new ParentChildScoreCollection(p, mutatedScores);
var bestOftheMutantStrains = initialBestCandidatesWithMutatedVariants
.SelectMany(x => x.Children)
.OrderByDescending(s => s.Score)
.First();
// Compute some stats for these mutations. These stats are for interest only, they don't affect any computation.
Benchmarking.Mutation.AddDataPoints(initialBestCandidatesWithMutatedVariants);
var rec = new Recogniser <TSymbol>(representedValue, bestOftheMutantStrains.Scorer);
RegisterRecogniser(rec);
}
/// <summary>
/// Calculates a score for the given input data. The higher the score, the more
/// likely that the input image represents the value of <c>Symbol</c>.
/// </summary>
public int Score(WrappedBitmap inputData)
{
if (inputData == null)
{
throw new ArgumentNullException(nameof(inputData));
}
return(_scorer.Score(inputData));
}
/// <summary>
/// Returns a list of symbols the classifier recognises, paired
/// with the score that the given image matches that symbol
/// </summary>
public Dictionary <TSymbol, int> Classify(WrappedBitmap input)
{
return(_recognisers
.Select(r => new
{
Symbol = r.Key,
Score = r.Value.Score(input)
})
.ToDictionary(kvp => kvp.Symbol, kvp => kvp.Score));
}
/// <summary>
/// Calculates a score for the given input data. The higher the score, the more
/// likely that the input image represents the value of <c>Symbol</c>.
/// </summary>
public int Score(WrappedBitmap inputData)
{
int score = 0;
foreach (var point in _pixelsToSample)
{
if (inputData.IsPixelSet(point.X, point.Y))
{
score += _tuningParams.PositiveScoreOffset;
}
else
{
score -= _tuningParams.NegativeScoreOffset;
}
}
return(score);
}
static void Main(string[] args)
{
WrappedBitmap a1w = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\a1.png");
WrappedBitmap a2w = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\a2.png");
WrappedBitmap a3w = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\a3.png");
//WrappedBitmap bw = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\b.png");
WrappedBitmap cw = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\c.png");
//WrappedBitmap dw = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\d.png");
//WrappedBitmap ew = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\e.png");
WrappedBitmap fw = WrappedBitmap.FromFile(@"..\..\..\..\sample-data\f.png");
var charClassifier_1 = new Classifier <char>();
charClassifier_1.Train(a1w, 'a', TuningParams.TrainingRounds);
TrainSampleCharacters(charClassifier_1);
var x = charClassifier_1.Classify(cw).OrderByDescending(kvp => kvp.Value).Select(kvp => kvp.Key);
Console.WriteLine("Input (C) is likely to be these letters (most likely first): {0}", string.Join(", ", x.Take(4)));
var y = charClassifier_1.Classify(fw).OrderByDescending(kvp => kvp.Value).Select(kvp => kvp.Key);
Console.WriteLine("Input (F) is likely to be these letters (most likely first): {0}", string.Join(", ", y.Take(4)));
var z = charClassifier_1.Classify(a1w).OrderByDescending(kvp => kvp.Value).Select(kvp => kvp.Key);
Console.WriteLine("Input (A) is likely to be these letters (most likely first): {0}", string.Join(", ", z.Take(4)));
var z1 = charClassifier_1.Classify(a2w).OrderByDescending(kvp => kvp.Value).Select(kvp => kvp.Key);
Console.WriteLine("Input (A2) is likely to be these letters (most likely first): {0}", string.Join(", ", z1.Take(4)));
var z2 = charClassifier_1.Classify(a3w).OrderByDescending(kvp => kvp.Value).Select(kvp => kvp.Key);
Console.WriteLine("Input (A3) is likely to be these letters (most likely first): {0}", string.Join(", ", z2.Take(4)));
//new Recogniser<char>('A').DrawSamplePoints().Save(@"..\..\..\..\sample-data\hello world.bmp");
Console.WriteLine("Press any key to quit");
Console.ReadKey();
}