Google OCR for Python 3

Uses Google Cloud Vision to perform Optical Character Recognition on jpg and png images.

Usage

1. Clone this repo

git clone https://github.com/ammo0110/Google-OCR

2. Install the requirements

pip3 install -r requirements.txt

3. Enable the Cloud Vision API from Google Cloud Platform Console. Refer to this
4. Get an API key for yourself. Refer to this
5. Once you have the API key, execute main.py file with following arguments

python3 main.py `path_to_api_key_file` `path_to_input_image`

Other features

1. -o/--output flag for redirecting output to a file
2. Recursive mode for recognizing a complete directory of jpg/png images at once
3. Multithreaded processing in case of recursive mode

For help, use

python3 main.py --help

Afterthoughts

Google provides two kinds of APIs for its GCP services: REST APIs and the other language specific APIs. I immediately found some issues with the Python specific APIs, which are:

1. They are written only for Python 2.
2. You have to install Google Libraries on your system
3. Unlike REST APIs, the authetication process for using these APIs is not decoupled from the API library itself. Although easier to use, I don't think that this kind of approach is practical from a design point of view.

Anyways, I decided to give it a try and wrote a program for Speech Recognition using the Python specific APIs. Then I realized one more problem with it. In case of an internet connection failure, these APIs don't report any error.

So now I have concluded that for Python, using REST APIs is the only option since it solves all the aforementioned problems.