CImg VS tesseract-ocr

CImg : http://cimg.eu/ (C++)

Then it uses tesseract to analyze each image, extracting the text.

Here is a little and unspectacular script to read text from screen. It uses tesseract to ocr the text and import command from ImageMagick to make a screenshot. Then the script outputs the recognized text to stdout. You could replace the screenshot tool with something else you like, but the script expects the created files.

The only open source OCR I know is Tesseract

Ran the first open source command line time for OCR that I could find, in this case https://github.com/tesseract-ocr/tesseract .. the command was pretty straight forward: tesseract -l eng book.tif out_from_tiff Again.. a simple shell script should be easy enough to write and apply it to all pages. The output did have a form feed character at the bottom.. Obviously you can manually delete it but that would take forever.. so simply run..

That being said, the tool is using a tesseract-related wrapper, and it might be having trouble picking up the text.

compiling Tesseract OCR or getting the binary https://github.com/tesseract-ocr/tesseract and downloading it locally to your project

I'm going to use the Tesseract OCR engine and library, and its Python wrapper PyTesseract for text extraction. But there are numerous libraries out there to extract text from an image. In a real world application I would probably use cloud services from AWS, Google or Microsoft to handle this task.

If this doesn’t fix it then check out this GitHub issue for more troubleshooting steps.

If you want an SDK for this you can use https://github.com/PaddlePaddle/PaddleOCR or https://github.com/tesseract-ocr/tesseract

This script uses the tesseract-ocr engine and some pip libraries. I've made it to be as user-friendly as I could and (theoretically) could translate from and to any language. It works with any PDF file, whether it is generated with any word proccessing software (MS Word, libreoffice writer...) or from a scanned document.