How can I add Arabic as a recognition language to Lios?

[Linux for blind general discussion]

Hi Ibrahim,

Not sure where the issue is.

To investigate, could you please send provide a link to a pdf in Arabic (or send
me privately the pdf itself) that we could both try?

Thanks

Cheers,
Didier

Le 20/12/2021 à 05:09, Linux for blind general discussion a écrit :
> Thanks Didier,
> 
> Unfortunately, I tried your suggestion and still got zero accuracy of OCR of
> Arabic letters. I typed a Paragraph of Arabic and printed and then scanned, but
> the outcome was gibbirish, all in latin characters. I typed a paragraph in
> English and printed it out, when I scanned, the OCR accuracy was excellent.  So,
> ther is some problem with the Arabic recognition.  It seems to me that although
> Arabic is listed, the OCR engine is not actually trying to recognize Arabic.
> 
> Cheers,
> 
> Ibrahim
> 
> On 12/17/21 6:48 PM, Linux for blind general discussion wrote:
>> Hi Ibrahim,
>>
>> You do not need to add anything special, the files
>> /usr/share/tessdata/Arabic.traineddata being included in the package
>> tesseract-data in Slint.
>>
>> Juts open lios, then in menu select Preferences then Preferences recognition and
>> select:
>> Engine: Tesseract
>> Language: Arabic
>>
>> I don't have a scanner at hand but downloaded this file:
>> https://fada.birzeit.edu/bitstream/20.500.11889/6910/1/mkhaldi%20Sahar%20Research.pdf
>>
>> then I opened it in Lios (menu File then Open).
>>
>> The file was recognized and the text properly extracted.
>>
>> Copying a paragraph of the extracted text and pasting it in translate.google.fr
>> allowed me to read it in French <smile>
>>
>> Cheers,
>>
>> Didier
>>
>> Le 18/12/2021 à 00:10, Linux for blind general discussion a écrit :
>>> Hi All,
>>>
>>> This question is primarily to Didier:
>>>
>>> How can I add Arabic dictionary to Lios so that I can use my scanner to scan
>>> Arabic text? I assume I will also be able to run Arabic.pdf files through Lios
>>> and as such I will have access to a lot of Arabic books available on the net.
>>>
>>> Cheers,
>>>
>>> Ibrahim