Character confidence is the same for the whole word.

[fhbiii]

Hello,

I'm trying to get individual character confidences for each character on a page. Long story short, I need each word on a page, but I'd like to be able to mark characters with a specifically low confidence. I'm reading each character, and building collections based on a custom class.

My issue is that every character within a word is returning the same confidence value. It's a new value for each word, almost as if it is simply returning the averaged word confidence.

Thank you in advance for your assistance.
My vb.net code is below:

 Private Sub PageToICRWords(page As Tesseract.Page)   
    Dim IT As Tesseract.ResultIterator = page.GetIterator
    Dim Chars As New ICRCharacterCollection
    Dim CurrentRow As Integer = -1

    Do
        Do
            Do
                Do
                    Do
                        Try
                            If IT.IsAtBeginningOf(PageIteratorLevel.TextLine) Then CurrentRow += 1
                            If IT.IsAtBeginningOf(PageIteratorLevel.Symbol) Then
                                Dim MyChar As New ICRCharacter
                                MyChar.Value = IT.GetText(PageIteratorLevel.Symbol)
                                If MyChar.Value > "" Then
                                    MyChar.Confidence = IT.GetConfidence(PageIteratorLevel.Symbol)
                                    Dim TesRec As New Rect
                                    IT.TryGetBoundingBox(PageIteratorLevel.Symbol, TesRec)
                                    MyChar.Box = New Rectangle(TesRec.X1, TesRec.Y1, TesRec.Width, TesRec.Height)
                                    Chars.Add(MyChar)
                                End If
                            End If
                            If IT.IsAtFinalOf(PageIteratorLevel.Word, PageIteratorLevel.Symbol) Then
                                If Not Chars Is Nothing AndAlso Chars.Count > 0 Then
                                    Me.Add(New ICRWord(Chars, CurrentRow))
                                End If
                                Chars = New ICRCharacterCollection
                            End If
                        Catch ex As Exception
                            LogErrorLocal("PageToICRWords", ex)
                        End Try
                    Loop While IT.Next(PageIteratorLevel.Symbol)
                Loop While IT.Next(PageIteratorLevel.Word)
            Loop While IT.Next(PageIteratorLevel.TextLine)
        Loop While IT.Next(PageIteratorLevel.Para)
    Loop While IT.Next(PageIteratorLevel.Block)


End Sub

[charlesw]

You'll need to set 'save_blog_choices' to True.

Source: http://stackoverflow.com/questions/11098845/character-wise-confidence-values-using-tesseract-3-01

And just to be thorough an example too :)

using (var engine = new TesseractEngine(@"./tessdata", "eng", EngineMode.Default)) {
    engine.SetVariable("save_blob_choices", true);

    using (var img = Pix.LoadFromFile(testImagePath)) {
        using (var page = engine.Process(img)) {
            int symbolCount = 0;
            using (var iter = page.GetIterator()) {
                iter.Begin();
                do {
                    if (iter.IsAtBeginningOf(PageIteratorLevel.Block)) {
                        Console.WriteLine("START BLOCK");
                    }

                    do {
                        do {
                            do {
                                if (iter.IsAtBeginningOf(PageIteratorLevel.Word)) {
                                    Console.WriteLine("\nSTART WORD");
                                }

                                do {
                                    Console.WriteLine("symbol: {0}, confidence: {1}",
                                        iter.GetText(PageIteratorLevel.Symbol),
                                        iter.GetConfidence(PageIteratorLevel.Symbol));

                                    if (iter.IsAtFinalOf(PageIteratorLevel.Word, PageIteratorLevel.Symbol)) {
                                        Console.Write("END WORD");

                                    }
                                    symbolCount++;
                                } while (iter.Next(PageIteratorLevel.Word, PageIteratorLevel.Symbol));
                            } while (iter.Next(PageIteratorLevel.TextLine, PageIteratorLevel.Word));
                        } while (iter.Next(PageIteratorLevel.Para, PageIteratorLevel.TextLine));


                        if (iter.IsAtFinalOf(PageIteratorLevel.Block, PageIteratorLevel.Para)) {
                            Console.WriteLine("\nEND BLOCK");
                        }
                    } while (iter.Next(PageIteratorLevel.Block, PageIteratorLevel.Para));
                } while (iter.Next(PageIteratorLevel.Block));
            }

            Console.WriteLine("Confidence: {0}", page.GetMeanConfidence());
            Console.WriteLine("Symbol Count: {0}", symbolCount);
        }
    }
}

[fhbiii]

Charles, thank you so much for your response. Unfortunately while I believe that should work, I'm now getting: "Attempted to read or write protected memory. This is often an indication that other memory is corrupt."
When calling .GetConfidence(PageIteratorLevel.Symbol)

[fhbiii]

Charles, Playing with it a bit more, I've noticed that I only get this error when the character returned would be a blank space. In my code, if I change the line

 If MyChar.Value > "" Then

to

 If MyChar.Value.Trim > "" Then

I do not get the protected memory error.

This solution should work fine for me, so I will stop bugging you about it, I just wanted to keep you informed.
Thanks for your assistance.
-Fred

[rremani]

Hi,
Is there any way to get character confidence through the command line.
Thanks