Closed bodhisattwawiki closed 6 years ago
Installed this script in another computer with Ubuntu 14.04 and ran ocr for this test file. Still it is delivering empty text.
do_ocr_2018-02-25-02-35-19_log.txt all_text_for_Testocrbengali.djvu.txt
Same issue from my machine.
I have checked Bengali and Tamil book, its not export txt file properly, means empty txt file. Please fix this issue ASAP. Otherwise OCR job have stopped.
For Bengali Wikisource the pdf file is not working. If we use image file(png, jpg,jpeg) instead of PDF, Its working fine.
@tshrinivasan please create the script for JPG ans I will test for small PDF/DJVU file. When I use Imagemagick convert command in 50MB file its crash. but below 50MB its convert well. I am trying to here https://github.com/jayantanth/OCR4wikisource/blob/master/doocr.py , but its not.
I am also trying with imagemagick.
But, it crashes the system.
Trying with adding tmedelay for each page conversion.
Will update soon.
2018-03-16 14:01 GMT+05:30 Jayanta Nath notifications@github.com:
@tshrinivasan https://github.com/tshrinivasan please create the script for JPG ans I will test for small PDF/DJVU file. When I use Imagemagick convert command in 50MB file its crash. but below 50MB its convert well. I am trying to here https://github.com/jayantanth/ OCR4wikisource/blob/master/doocr.py , but its not.
— You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub https://github.com/tshrinivasan/OCR4wikisource/issues/99#issuecomment-373640257, or mute the thread https://github.com/notifications/unsubscribe-auth/ABNbOG2_wkOfFed3Gp9Wx586yoDply9vks5te3h2gaJpZM4SRyGV .
-- Regards, T.Shrinivasan
My Life with GNU/Linux : http://goinggnu.wordpress.com Free E-Magazine on Free Open Source Software in Tamil : http://kaniyam.com
Get Free Tamil Ebooks for Android, iOS, Kindle, Computer : http://FreeTamilEbooks.com
Hi @tshrinivasan could you please try
gs -q -DNOPAUSE -DBATCH -r400 -SDEVICE=a4 -sOutputFile=abcd%d.jpg abcd.pdf
Rate 20 page/Min but not system crashed of 200 mb file for me.
gs -q -DNOPAUSE -DBATCH -r400 -SDEVICE=a4 -sOutputFile=a%d.jpg a.pdf
Unknown device: a4
./base/gsicc_manage.c:1088: gsicc_open_search(): Could not find default_gray.icc
| ./base/gsicc_manage.c:1708: gsicc_set_device_profile(): cannot find device profile
Unrecoverable error: unknownerror in .special_op
Operand stack:
defaultdevice
Unrecoverable error: undefined in .uninstallpagedevice
Operand stack:
defaultdevice
Got above error.
But converting to png works
gs -q -DNOPAUSE -DBATCH -r400 -SDEVICE=png16m -sOutputFile=a%d.png a.pdf
this is working fine.
Will test more for performance and crashing.
Sorry the right code is below
gs -q -DNOPAUSE -DBATCH -r400 -SDEVICE=jpeg -sPAPERSIZE=a4 -sOutputFile=abcd%d.jpg abcd.pdf
Great.
This works nicely.
Will incorporate this to do_ocr.py
Any update?
Added the gs commands given to convert the pdf files to jpg files.
download the above file.
mv do_ocr_jpg.txt do_ocr_jpg.py
and run the do_ocr_jpg.py file and share the results.
use this file. fixed few spell errors.
mv do_ocr_jpg.py-v2.txt do_ocr_jpg-v2.py
and run the do_ocr_jpg-v2.py file and share the results.
with v2 version of script
Downloading the OCRed text
INFO:main:Running gdput.py -t ocr -f 1sEdaoQ2YBzXcckYFqK_JQYXYx_cFU1MW page_00118.jpg | tee page_00118.log sed: -e expression #1, char 12: unterminated `s' command
try this file. removed the feature of adding one more empty lines between paras. Let us fix it later.
mv do_ocr_jpg.py-v3.txt do_ocr_jpg-v3.py
and run the do_ocr_jpg-v3.py file and share the results.
Thanks @Shreeshrii for the tests. Found the OCR is not working for PDF and it works for JPG. In this issue, I am converting the PDF to JPG to get OCR running and to give results.
Check for the OCR gives text and not empty pages for the languages you know.
Once this is fixed, we can work on optimal file io.
Test v3 and share the CR results.
Thanks. I will try with new version.
Found the OCR is not working for PDF and it works for JPG.
Yes, I was getting OCR files of 3 bytes when converting with pdf and was wondering what was wrong, and found your posted solution here. Thanks for your prompt response.
Google is removing PDF support for OCR. Hence trying with JPG.
Let us work on one problem at a time.
Test the OCR workability in this issue.
Raise a new issue for filenames with space.
Thanks.
I have checked with files in Devanagari and Gujarati scripts. Conversion is happening from jpg.
With version 3, no sed error. Output file is greater than 3 bytes :)
Downloading the OCRed text
INFO:__main__:Running gdput.py -t ocr -f 1z5wcJsHDFYd16UQhs28U5cfUEbGV8JkN page_00002.jpg | tee page_00002.log
File location: /home/sanskrit/OCR4wikisource/page_00002.txt
File size in bytes: 3044
INFO:__main__:
Creating temp file touch page_00002.upload
INFO:__main__:
merged do_ocr_jpg.py-v3.txt with do_ocr.py and fixed this issue
OCR of this book delivered empty text. I am using Ubuntu 16.04. Same happened with Ravi's fork.