I'm looking for the best solution to get information out of +- 90.000 images which have similar markup. An online example I found which is similiar to my cases:


I need the information as raw data to be able to process them further.

For the numeric "round-formed" parts of the image I need to know the location of the numbers to be able to compare them.

I was thinking about OCR software. Any advice in the software to use for my specific case. Open source / freeware would be preferred.

  • Have you tried regular OCR to see if the numeric data and headers get recognized? I tried your sample image and it parsed the words and numbers fairly good on a free online upload/convert site. I'm thinking that if the numbers end up in the same positions in the resulting plain text file, you could simply parse them with some fancy shell scripting...
    – ivanivan
    Commented Oct 10, 2018 at 3:27

1 Answer 1


Using OpenCV & Python you can train a deep learning model to spot the text in images then Tesseract to OCR it - there is a tutorial that takes you through this step by step at https://www.pyimagesearch.com/2018/09/17/opencv-ocr-and-text-recognition-with-tesseract/ which is very well structured and includes the source code.

  • All Free, Gratis & Open Source
  • All cross platform.

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.