Open ZARk-be opened 1 month ago
It might just be too much for the platform API to process, though that might also depend on how you're providing the image to it. Try preprocessing the image to make the text as readable as possible: https://medium.com/technovators/survey-on-image-preprocessing-techniques-to-improve-ocr-accuracy-616ddb931b76
Hello
I'm trying to scan the MRZ ( Machine Readable Zone ) of an ID card. example here : [https://www.ibz.rrn.fgov.be/typo3temp/pics/476d8015a3.jpg](sample picture)
It's not very consistent, usually the first line is not getting picked up. Are there specific options i could try out ? I tried the SetTryHard but there's no real difference in my case.