Reworked distance info algo

Examples/Common/Distance/DistanceComparer.cs

@@ -26,11 +26,18 @@ public readonly struct DistanceComparer<T> : IDistanceComparer<T>
     Distance = Calculator.GetDistance(Reference, Hypothesis);
   }
 
+
   /// <inheritdoc />
-  public override string ToString() => Hypothesis switch
+  public override string ToString()
   {
-    null => "`null`",
+    var str = Hypothesis?.ToString();
-    var hyp when Equals(hyp, Reference) => Hypothesis.ToString() ?? string.Empty,
+
-    _ => $"<strong style='color: orange;' title='CER: {Distance}'>{Hypothesis}</strong>"
+    if (Hypothesis is var hyp && Equals(hyp, Reference))
-  };
+    {
+      return str ?? string.Empty;
+    }
+
+    return
+      $"<strong style='color: orange;' title='REf: {Reference}, CER: {Distance}'>{str ?? "-"}</strong>";
+  }
 }

Examples/ReportGenerator/Generator/Generator/HtmlDocumentGenerator.cs

@@ -80,39 +80,8 @@ public class HtmlDocumentGenerator : DocumentGeneratorBase
     AppendParagraph(text, default);
 
   /// <inheritdoc />
-  public override IDocumentGenerator AppendHeading(int level, string text)
+  public override IDocumentGenerator AppendHeading(int level, string text) =>
-  {
+    Append($"<h{level} id=\"{text}\">{text}</h{level}>");
-    if (_sectionLevel > 0)
-    {
-      // todo ??
-    }
-
-    var delta = level - _sectionLevel;
-    switch (delta)
-    {
-      case > 0:
-        for (int i = 0; i < delta; i++)
-        {
-          Append("<div>");
-        }
-
-        break;
-      case < 0:
-        for (int i = delta; i < 0; i++)
-        {
-          Append("</div>");
-        }
-
-        break;
-    }
-
-    Append("</div>");
-    Append("<div>");
-
-    _sectionLevel = level;
-
-    return Append($"<h{level}>{text}</h{level}>");
-  }
 
   /// <inheritdoc />
   protected override ITableGenerator MakeTable(int columns, Stream stream) =>

Examples/ReportGenerator/Models/ProcessorStat.cs

@@ -46,50 +46,40 @@ public readonly struct ProcessorStat : IDistanceComparer<IEnumerable<string>>
       hypothesis.OrderBy(s => s).ToArray()
     ) / reference.Count;
 
-    Words = reference.Select(r => GetDistanceInfo(r, hypothesis)).ToArray();
+    Words = GetDistanceInfos(reference, hypothesis).ToArray();
+    // Words = reference.Select(r => GetDistanceInfo(r, hypothesis)).ToArray();
   }
 
-  private static ICollection<IDistanceComparer<string>> GetDistanceInfos(
-    ICollection<string> reference,
-    ICollection<string> hypothesis
-  )
-  {
-    // todo avoid matching the same reference with a value multiple times
-    throw new NotImplementedException();
-  }
 
   /// <summary>
-  /// Compares the <paramref name="reference"/> with all given <paramref name="values"/>
+  /// Compares the <paramref name="referenceCollection"/> with all given values in the
-  /// and determines the <see cref="IDistanceComparer{T}"/> with the lowest error
+  /// <paramref name="hypothesisCollection"/>, determining the <see cref="IDistanceComparer{T}"/>
+  /// with the lowest error
   /// </summary>
-  private static IDistanceComparer<string> GetDistanceInfo(
+  private static IEnumerable<IDistanceComparer<string>> GetDistanceInfos(
-    string reference,
+    ICollection<string> referenceCollection,
-    IEnumerable<string> values
+    ICollection<string> hypothesisCollection
   )
   {
-    var result = new DistanceComparer<string>(reference);
+    var results = new List<IDistanceComparer<string>>();
 
-    // Determine character stat with lowest error
+    foreach (var reference in referenceCollection)
-    foreach (var value in values)
     {
-      // todo avoid matching the same reference with a value multiple times
+      var tResults = hypothesisCollection
+        .Select(hypothesis => new DistanceComparer<string>(reference, hypothesis))
+        .Cast<IDistanceComparer<string>>()
+        .ToList();
 
-      var stat = new DistanceComparer<string>(reference, value);
+      results.AddRange(tResults);
-      if (stat.Distance > result.Distance || (stat.Distance / reference.Length) > 0.6d)
+    }
+
+    var lookup = results
+      .OrderBy(result => result.Distance)
+      .ToLookup(result => result.Reference);
+
+    foreach (var reference in referenceCollection)
     {
-        // todo fine-tune threshold
+      yield return lookup[reference].FirstOrDefault() ?? new DistanceComparer<string>(reference);
-        continue;
-      }
-
-      result = stat;
-
-      if (stat.Distance <= 0)
-      {
-        // We cannot go lower than zero, break
-        return result;
     }
   }
-
-    return result;
-  }
 }

Examples/ReportGenerator/Program.cs

@@ -1,4 +1,5 @@
-using ReportGenerator.Generator.Generator;
+using Common.Extensions;
+using ReportGenerator.Generator.Generator;
 using ReportGenerator.Models;
 
 namespace ReportGenerator;
@@ -24,7 +25,8 @@ internal static class Program
     using var report = new ReportGenerator("OCR Report", document, scans)
       .AddComparison("Processing summary (Average)", v => v.Average())
       // .AddComparison("Processing summary (Cumulative)", v => v.Sum())
-      // .AddComparison("Processing summary (Median)", v => v.Median())
+      .AddComparison("Processing summary (Median)", v => v.Median())
+      // .AddProcessorStats("Processor Stats")
       .AddImageStatsFull("Scan Results");
 
     Console.WriteLine($"Saved report to '{path}'");

Examples/ReportGenerator/ReportGenerator.cs

@@ -39,7 +39,55 @@ public class ReportGenerator : IDisposable
   {
     Document.AppendHeading(2, title);
 
-    // todo show best/worst images per processor
+    var processors = new Dictionary<string, ICollection<ImageStats>>();
+    foreach (var image in Images)
+    {
+      foreach (var processor in image.Processors)
+      {
+        if (processors.TryGetValue(processor.Name, out var images))
+        {
+          images.Add(image);
+        }
+        else
+        {
+          processors.Add(processor.Name, new List<ImageStats> { image });
+        }
+      }
+    }
+
+    foreach (var (processor, images) in processors)
+    {
+      var ordered = images
+        .Select(i => (Stats: i, Distance: i
+            .Processors
+            .Where(p => p.Name.Equals(processor))
+            .Select(p => p.Distance)
+            .Average()
+          ))
+        .OrderBy(i => i.Distance)
+        .ToArray();
+
+      Document
+        .AppendHeading(3, processor)
+        .AppendTable(2, table =>
+          {
+            table.AppendHeader(new[] { "Image", "Preview", "Distance" });
+
+            foreach (var (stats, distance) in ordered)
+            {
+              var imgPath = Path.Combine("results", $"{processor}.00.{stats.ImageName}.png");
+
+              table.AppendRow(new[]
+              {
+                $"<a href=\"#{stats.ImageName}\">{stats.ImageName}</a>",
+                Document.FormatImage(imgPath, new Bounds(0, 150)),
+                distance.ToString("F2")
+              });
+            }
+          }
+        );
+    }
+
     return this;
   }
 
@@ -75,10 +123,15 @@ public class ReportGenerator : IDisposable
             {
               var imgPath = Path.Combine("results", $"{processor.Name}.00.{stat.ImageName}.png");
 
+              string summarizedCer = string.Empty;
+              summarizedCer = processor.Words
+                .Average(s => s.Distance)
+                .ToString("F2");
+
               table.AppendRow(processor.Words
                 .Select(s => s.ToString() ?? string.Empty)
                 .Prepend(Document.FormatImage(imgPath, new Bounds(0, 150)))
-                .Prepend(processor.Words.Average(s => s.Distance).ToString("F2"))
+                .Prepend(summarizedCer)
                 .Prepend($"{processor.Distance * 100:F1}%")
                 .Prepend($"{processor.ProcessingTime * 1000:F1}ms")
                 .Prepend(processor.Name)