A simple problem with OCR in .NET C #

Question

I do some OCR things and the screen creaks. As a result, I have many files that look like this.

All I need to do is a very simple OCR in C # in these files. I pulled my hair, trying to get different libraries to work (Tessnet2, Puma, MODI), and they had many different problems, forcing them to even start from C #.

What do you guys recommend for something so simple?

Thanks!

Answer 1

OCR programs are not designed to read low-resolution images. Even some of the top best commercial OCR engines have difficulty reading screenshots.

Tesseract needs good, clean images, even under normal conditions, to get decent results. There could be several reasons why you get poor results. If you post sample images and output results, we can better explain the results. Problems include color backgrounds, text zoning errors, small characters, artifacts ....

Tesseract seems to get much better results if you train it with the fonts you want to read.

Answer 2

There is a web interface for OCR that you can try, here is an example of C # how to use it: http://snipt.org/lOgh/ (you first need to register the API key in http://www.wisetrend.com /wisetrend_ocr_cloud.shtml - search for the "Register for free" button).

Disclaimer: WiseTrend is a client of my company.