Farsi numbers mismatch order

[مریم طوطی‎]

I installed tesseract ocr on my project, but when I test Persian numbers the result is incorrect. As sample
۳۴۵۶۷ will gives me
۷۳۴۵۶
Or
۱۲۳۴۵۶۷۸۹ will be
۹۸۷۴۵۶۳۲۱
Each test has different result.
My project is on dead line. Please give me a solution. Thanks.

[namv...@gmail.com]

you can try ara.traineddata or fas.traineddata from here:

https://github.com/tesseract-ocr/tessdata

there is also traineddata for persian in this page:

https://github.com/reza1615/PersianOcr

if you create your own traindata you can use convert unicharset to RTL.py in reza1615 page. (run this code after create the unicharset file)