No recognizing Arabic numbers, but recognizes letters

[Aijolomohi Egwaikhide]

Hi, I am working with OCR to recognize arabic words and numbers(dates) from a scanned pdf (I have done some enhancing on it), but it cant seem to accurately read arabic numbers but it reads the letters properly. what can i do to make this better?

An example of a numbers (in date format) 

The first one is the input, the second one is the output and it read it as English even though i specify that it is all arabic - when I put the document in google docs and read as a word document, it reads it fine