You have experience in training tesseract. I want to OCR Identity Cards which uses different fonts on each row.
You will take sample Scanned Cards, isolate the different fonts used within them, and run enough training on samples of each that accuracy will be improved on future Cards, and return me the tessdata folder, along with whatever else I need to run tesseract.
If it matters, this will be an Windows installation of tesseract 3.02 with C# Wrapper
I can provide you with unlimited samples, you don't have to train on all pages of them, of course, but you do need to go through a sampling and identify and isolate the crucial fonts, and train on enough to improve accuracy.
1. the cards contains both Arabic and English text with numbers, I m interested in Arabic Fonts.
2. Fonts and Sample will be delivered once approval.
3. Arabic Characters has different forms based on the location of the work( start, between, end, isolated) I will help you to identify different...