henrivain / tesseractocrmaui Goto Github PK

Will you upgrade this library to use both Andorid & iOS?
Thank you

Problem

User cannot configure Tesseract parameters of the injected ITesseract api and must use underlying TessEngine api.
Because of this dependency injection cannot be used when user wants to change for example tessedit_char_whitelist or tessedit_char_blacklist.

Current way of doing

string traineddataFolder = FileSystem.Current.CacheDirectory;

// Load data
var tranineddataPath = Path.Combine(traineddataFolder , "eng.traineddata");
if (!File.Exists(tranineddataPath)) 
{
    using Stream traineddata = await FileSystem.OpenAppPackageFileAsync("eng.traineddata");
    FileStream fileStream = File.Create(tranineddataPath);
    traineddata.CopyTo(fileStream);
}

// Create and configure engine
using var engine = new TessEngine("eng", traineddataFolder );
bool success = engine.SetVariable("tessedit_char_whitelist", "mychars");

// Recognize text
using var image = Pix.LoadFromFile(@"\path\to\file.png");
using var result = engine.ProcessImage(image);
string text = result.GetText();

Suggested fix

Addition to ITesseract API

Add optional configuration Property to pass into ITesseract

public interface ITesseract
{
    ...

    Action<TessEngine> EngineConfiguration { get; set; }
}

Changes in Tesseract.cs

Method internal RecognizionResult Recognize(Pix pix, string tessDataFolder, string[] traineddataFileNames)

Change

try
{
    using var engine = new TessEngine(languages, tessDataFolder, Logger);
    using var page = engine.ProcessImage(pix);
    
    confidence = page.GetConfidence();
    text = page.GetText();
}
...

To

try
{
    using var engine = new TessEngine(languages, tessDataFolder, Logger);
    // Configure
    EngineConfiguration(engine);

    using var page = engine.ProcessImage(pix);
    confidence = page.GetConfidence();
    text = page.GetText();
}
...

Intended use

Set configuration func before running process

Tesseract.EngineConfiguration = engine => 
{
    // These characters are not recognized
    engine.SetVariable("tessedit_char_blacklist", "bad");
}
var result = await Tesseract.RecognizeTextAsync(@"my\image\path.png");

Output bin folder does not have lib -folder inside

Also can't find them by searching

With project reference everything works fine, but with nuget package dll cannot be loaded.

Android works fine.

Affected platforms: Windows

Hi,

Thank you for making this. Would you be able to add text whitelist capability?

Hi,

The following jpeg images are not able to be read:

The following does work so I'm guessing it's unable to read single digits.

I tested in windows TesseractOcrMauiTestApp project along with an android app. For android, I do know Xamarin.Tesseract package is able to read them properly but I am attempting to phase it out.

Pix.LoadFromMemory(byte[]) does not work on Android with jpeg images

• Leptonica cannot load jpeg images from memory as byte array.
• Loading image from path or loading png image works fine.

Possible explanation

• Android native library might not have all needed 3rd party libs installed
• Openjpg library is not currently built into android native libs
• Leptonica bug

Affected Platforms:

Android physical devices

Reproduction example

// Assumptions
// Tessdata already loaded 

byte[] bytes = File.ReadAllBytes(@"my\image\path.jpg");

// This throws "System.IO.IOException: 'Failed to load image from memory.'"
// Meaning Leptonica returns IntPtr.Zero
using Pix image = Pix.LoadFromMemory(bytes);

Workarounds

• Load saved image from device storage using

Pix.LoadFromFile(@"my\image\path.jpg");

• Convert image to png -format

These do not completely fix the need to pass jpeg images straight from memory to tesseract without changing them before

Hi @henrivain
I think its great you started working on this iOS-branch - My project is for Android and later on for iPhone.
I have tried using the Tesseract-nuget (Not yours) and it work fine for my project but it only works in Windows.
For my project i found that running Tesseract with enginemode: EngineMode.TesseractOnly was better than default and Lstm.
Is it possible to add EngineMode (TesseractOnly, LstmOnly, TesseractAndLstm and Default)to your tesseractOcrMaui.EngineConfiguration?
Regards

Originally posted by @ThomasMunch in #11 (comment)