Guten OCR

To use Guten OCR in a web browser, install @gutenye/ocr-browser. When calling Ocr.create(), you must provide a models configuration object specifying the paths to the required ONNX model files and the dictionary file.

bun add @gutenye/ocr-browser
import Ocr from '@gutenye/ocr-browser'
const ocr = await Ocr.create({
  models: {
    detectionPath: '/assets/ch_PP-OCRv4_det_infer.onnx',
    recognitionPath: '/assets/ch_PP-OCRv4_rec_infer.onnx',
    dictionaryPath: '/assets/ppocr_keys_v1.txt'
  }
})
const result = await ocr.detect('/a.jpg')

To run the C++ example, you must install the ONNX Runtime and OpenCV dependencies, then use the provided build tool to start the project.

1. Download the appropriate onnxruntime release from the Microsoft ONNX Runtime releases page.
2. Install OpenCV via Homebrew.
3. Run the project using ./ake start.

download onnxruntime from https://github.com/microsoft/onnxruntime/releases
brew install opencv
./ake start

To set up and run the React Native example project, use bun for dependency installation, make start to launch the bundler, and then build the application using Xcode for iOS devices/simulators.

bun install
make start
# Then use Xcode to build the app

For C++ environments, include native-ocr.h and instantiate a NativeOcr object to perform detection.

#include "native-ocr.h"
NativeOcr* ocr = new NativeOcr(..)
auto result = ocr->detect("a.jpg");

To set up the React Native package, navigate to the package directory and copy the required assets into a resources folder.

Important Limitation: The iOS Simulator is not supported on Apple Silicon (M1/M2/M3) because ExpoModulesCore can only run on the x86_64 simulator via Rosetta. You must use a physical device or an x86_64 simulator environment.

cd packages/react-native
cp -rL ../../assets resources

To develop on this repository, you must have Git LFS installed to properly clone the large model assets.

brew install git-lfs 
git clone git@github.com:gutenye/ocr.git

To use Guten OCR in a Node.js environment, install the @gutenye/ocr-node package. You can initialize the OCR engine using Ocr.create() and then perform text detection on an image file path using the .detect() method.

bun add @gutenye/ocr-node
import Ocr from '@gutenye/ocr-node'
const ocr = await Ocr.create()
const result = await ocr.detect('a.jpg')

To use Guten OCR in a React Native project, install @gutenye/ocr-react-native. The engine can be initialized and used similarly to the Node.js implementation.

bun add @gutenye/ocr-react-native
import Ocr from '@gutenye/ocr-react-native'
const ocr = await Ocr.create()
const result = await ocr.detect('a.jpg')

To run the browser-based example project, you must first install the dependencies for the entire monorepo and then install the local dependencies for the example directory. Finally, use the ake build tool to start the development server.

Ensure you are in the correct directory relative to the browser/example folder when running these commands.

cd ../.. && bun install
cd browser/example
bun install
./ake start

To run the Node.js example, you must first install the dependencies for the entire repository (monorepo) and then install the dependencies specific to the examples/node directory. Finally, use the ake command to start the example with a specific image asset.

Ensure you are in the examples/node directory before running the local installation and start commands.

# From the examples/node directory:
cd ../.. && bun install
cd examples/node
bun install
./ake start assets/cn-01.jpg

The @gutenye/ocr-browser package provides a browser-compatible implementation of the Guten OCR API. It automatically registers the necessary browser-specific backends (such as onnxruntime-web for inference and FileUtils/ImageRaw for image handling) so you can use the standard Ocr interface.

To use it, import the default Ocr export from the package. This allows you to perform OCR tasks directly in web environments using the same API surface as the Node.js or React Native versions.

import Ocr from '@gutenye/ocr-browser';

// Use the Ocr instance as you would in any other environment
// const results = await Ocr.recognize(imageSource);

If you encounter errors during execution, check the following common dependency issues:

• Missing libavcodec.60.dylib: Reinstall ffmpeg@6 using Homebrew.
• OpenCV not found: Reinstall opencv using Homebrew.

- libavcodec.60.dylib not found: brew reinstall ffmpeg@6
- OpenCV not found: brew resinstall opencv