epassaro / scr2ocr

Screenshot to OCR utility

scr2ocr

Screenshot to OCR utility

Installation and usage

Prerequisites

GNU/Linux

It should work on any distribution with xclip installed. Wayland is not supported (see: https://github.com/python-pillow/Pillow/issues/6312)]

1. Download the zip file from the releases section

$ wget https://github.com/epassaro/scr2ocr/releases/download/latest/scr2ocr-linux-64.zip

2. Extract and execute the src2ocr binary file

unzip scr2ocr-linux-64.zip
cd scr2ocr
./scr2ocr

⚠️ NOTE: Loading the JamSpell model for the first time could take a while, don't worry!

3. Resize the window to cover the area you want to capture and press the square button
4. Captured text should be automatically copied to your clipboard 🎉

macOS

Currently not tested. Not built. Use development install.

Windows

Currently not tested, JamSpell is not supported. Please do not open issues about the Windows build!

Development

1. Requires installing Miniconda or a Miniforge variant (we recommend Mambaforge)
2. Install the required packages in a new virtual environment using the lockfile meant for your platform

$ mamba create -n mamba create -n scr2ocr --file conda-linux-64.lock

3. Activate the environment and install jamspell with pip

$ mamba activate scr2ocr
$ pip install jamspell==0.0.12

4. Place a JamSpell model file named model.bin inside the jamspell folder
5. Run the script

$ python scr2ocr.py

Authors

• @epassaro
• @agostinaf
• @fitosb

License

scr2ocr's Issues

Make `scr2ocr` multilingual

Currently is intended to use only with text in Spanish (Tesseract and JamSpell)

Tweak `tesseract` flags via UI

See: https://tesseract-ocr.github.io/tessdoc/ImproveQuality