Tesseract 4 not reading Arabic numbers accurately using custom trained data file

[Mobeen Ali]

Current Behavior:

I've followed the wiki and details given in the wiki/Training Tesseract - 4.00. There were no errors in creation of the traineddata file. 

I wanted to create my own ara_custom.traineddata file specifically to read dates in arabic, so it has "٠١٢٣٤٥٦٧٨٩" (0-9 numeric characters in arabic) with a "/" forward slash only.

The format for arabic date is:

٢٠١٩/٠٩/٢٥

yyyy/mm/dd

My ara.training_text file is: attached as ara.training_text.txt (for uploading only else i use the file without txt extension)

My ara.wordlist file is:  attached as ara.wordlist.txt (for uploading only else i use the file without txt extension)

Text in image: ٢٠٠٩/١١/١٢ (32.jpg)

Tesseract reads: ٢٤٠٩/١١/١٢ (32.txt)

Text in image: ١٩٧٩/٠١/٢٨ (24.jpg)

Tesseract reads: ١٦٩٧٦    //٠١//٧٢٨ (24.txt)

Text in image: ٢٠١٥/١١/٢٢ (12.jpg)

Tesseract reads: ٢٠١٥/١١/٧٢ (12.txt)

What i observed is I've issue in my training_text file. I've attached the file above. Please guide me for this error as i have failed to find any solution myself.

P.s. I've studied the Hallucination effect also which is given in the wiki and tried to implement it as i understood, but no luck.

[Béchir Gmati]

hi plz i have this error when i execute the command line of combine-lang-model  how i can fix it

 --

   GMATI Béchir

   Élève Ingénieur Business Intelligence & Big Data

  

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.
To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-oc...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/6878322a-c591-481b-b0d8-0befd76cbd22%40googlegroups.com.

[Shree Devi Kumar]

You are missing https://github.com/tesseract-ocr/langdata_lstm/blob/master/radical-stroke.txt

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAKGr270sh9rw1qzix%3Dhqgf3PwdGy8jDavrrBeCAnSUuNFdmrFw%40mail.gmail.com.

--

____________________________________________________________
भजन - कीर्तन - आरती @ http://bhajans.ramparivar.com

[nourhan magdy]

how can i use this text file? i downloaded ara folder and coppied it to my tessdata but it didnt work

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/6878322a-c591-481b-b0d8-0befd76cbd22%40googlegroups.com.

--
You received this message because you are subscribed to the Google Groups "tesseract-ocr" group.

To unsubscribe from this group and stop receiving emails from it, send an email to tesser...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/CAKGr270sh9rw1qzix%3Dhqgf3PwdGy8jDavrrBeCAnSUuNFdmrFw%40mail.gmail.com.

[Piyush Chandra]

You need to put the radical stroke file in your script_dir folder.

[nourhan magdy]

It's a text file not trained, I put it in the scripts file but it didn't work
Returns empty
And the same with ara characters returns empty or incorrect answer

To unsubscribe from this group and stop receiving emails from it, send an email to tesseract-ocr+unsubscribe@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/tesseract-ocr/7064e149-0f32-4072-8f50-9101ba341a51%40googlegroups.com.

[write2...@gmail.com]

Do anyone still have Arabic date traineddata ?

pls share the tesseract weight files if anyone able to extract attached example img.

date format:

٢٠١٩/٠٩/٢٥