Tesseract4Android

Fork of tess-two rewritten from scratch to build with CMake and support latest Android Studio and Tesseract OCR.

The Java/JNI wrapper files and tests for Leptonica / Tesseract are based on the tess-two project, which is based on Tesseract Tools for Android.

Dependencies

This project uses additional libraries (with their own specific licenses):

• Tesseract OCR 5.5.0
• Leptonica 1.85.0
• libjpeg v9f
• libpng 1.6.44

Prerequisites

• Android 5.0 (API 21) or higher
• A v4.0.0 trained data file(s) for language(s) you want to use.
  • These files must be placed in the (sub)directory named tessdata and the path must be readable by the app. When targeting API >=29, only suitable places for this are app's private directories (like context.getFilesDir() or context.getExternalFilesDir()).

Variants

This library is available in two variants.

• Standard - Single-threaded. Best for single-core processors or when using multiple Tesseract instances in parallel.
• OpenMP - Multi-threaded. Provides better performance on multi-core processors when using only single instance of Tesseract.

Usage

You can get compiled version of Tesseract4Android from JitPack.io.

1. Add the JitPack repository to your project root build.gradle file at the end of repositories:

allprojects {
    repositories {
        ...
        maven { url 'https://jitpack.io' }
    }
}

2. Add the dependency to your app module build.gradle file:

dependencies {
    // To use Standard variant:
    implementation 'cz.adaptech.tesseract4android:tesseract4android:4.8.0'

    // To use OpenMP variant:
    implementation 'cz.adaptech.tesseract4android:tesseract4android-openmp:4.8.0'
}

3. Use the TessBaseAPI class in your code:

This is the simplest example you can have. In this case TessBaseAPI is always created, used to recognize the image and then destroyed. Better would be to create and initialize the instance only once and use it to recognize multiple images instead. Look at the sample project for such usage, additionally with progress notifications and a way to stop the ongoing processing.

// Create TessBaseAPI instance (this internally creates the native Tesseract instance)
TessBaseAPI tess = new TessBaseAPI();

// NOTE: TessBaseAPI is not thread-safe. If you want to process multiple images in parallel,
// create separate instance of TessBaseAPI for each thread.

// Given path must contain subdirectory `tessdata` where are `*.traineddata` language files
// The path must be directly readable by the app
String dataPath = new File(context.getFilesDir(), "tesseract").getAbsolutePath();

// Initialize API for specified language
// (can be called multiple times during Tesseract lifetime)
if (!tess.init(dataPath, "eng")) { // could be multiple languages, like "eng+deu+fra"
    // Error initializing Tesseract (wrong/inaccessible data path or not existing language file(s))
    // Release the native Tesseract instance
    tess.recycle();
    return;
}

// Load the image (file path, Bitmap, Pix...)
// (can be called multiple times during Tesseract lifetime)
tess.setImage(image);

// Start the recognition (if not done for this image yet) and retrieve the result
// (can be called multiple times during Tesseract lifetime)
String text = tess.getUTF8Text();

// Release the native Tesseract instance when you don't want to use it anymore
// After this call, no method can be called on this TessBaseAPI instance
tess.recycle();

Sample app

There is example application in the sample directory. It shows basic usage of the TessBaseAPI inside ViewModel, showing progress indication, allowing stopping the processing, filtering results based on confidence, and more.

It uses sample image and english traineddata, which are extracted from the assets in the APK to app's private directory on device. This is simple, but you are keeping 2 instances of the data file (first is kept in the APK file itself, second is kept on the storage) - wasting some space. If you plan to use multiple traineddata files, it would be better to download them directly from the internet rather than distributing them within the APK.

Building

You can use Android Studio to open the project and build the AAR. Or you can use gradlew from command line.

To build the release version of the library, use task tesseract4android:assembleRelease. After successful build, you will have resulting AAR files in the <project dir>/tesseract4Android/build/outputs/aar/ directory.

Or you can publish the AAR directly to your local maven repository, by using task tesseract4android:publishToMavenLocal. After successful build, you can consume your library as any other maven dependency. Just make sure to add mavenLocal() repository in repositories {} block in your project's build.gradle file.

Android Studio

• Open this project in Android Studio.
• Open Gradle panel, expand Tesseract4Android / :tesseract4Android / Tasks / other and run assembleRelease (to get AAR).
• Or in the same panel expand Tesseract4Android / :tesseract4Android / Tasks / publishing and run publishToMavenLocal (to publish AAR).

GradleW

• In project directory create local.properties file containing:

sdk.dir=c\:\\your\\path\\to\\android\\sdk
ndk.dir=c\:\\your\\path\\to\\android\\ndk

Note for paths on Windows you must use \ to escape some special characters, as in example above.

• Call gradlew tesseract4android:assembleRelease from command line (to get AAR).
• Or call gradlew tesseract4android:publishToMavenLocal from command line (to publish AAR).

License

Copyright 2019 Adaptech s.r.o., Robert Pösel

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

    http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.