gunthercox/ocr-service

A stand alone character recognition micro-service with a RESTful API

OCR Service

This is a stand alone character recognition microservice that can be accessed via a REST API.

The service's expects json in the following format:

{
    'image': '<image byte content>'
}

Data will be returned in the following format:

{
    "text": [
        "Line of text one",
        "Line of text two"
    ]
}

Running the container

This application is fully Dockerized and can be started with docker-compose.

docker-compose up

ocr-service's Issues

Support for additional data types in post request

Base64 encoded images aren't the most efficient way to transmit images.

Update pytesseract version

Version 0.2 has been released. The requirements file for this service was previously pinned to a commit to get the latest functionality from pytesseract.

https://github.com/madmaze/pytesseract/releases

Handwriting tests

I'm not sure if tesseract supports this now, but hand writing recognition would be cool.

Image orientation correction

I'm not sure if this matters yet, but I think most OCR libraries make the assumption that the text is already oriented the right way.

Add tests for different languages