Discussion:
[Wikisource-l] Zero output for OCR using Google Drive API
Bodhisattwa Mandal
2018-03-15 12:00:36 UTC
Permalink
Hi,

For last few weeks, Google OCR using Drive API has stopped working for few
Indic scripts like Bengali and Devanagari affecting Bengali, Sanskrit and
Assamese Wikisource. It's still working for other Indic scripts, I guess.

If anyone have any contact with Google, we need to know if this temporary
or any plan for major change is underway.

This is extremely important as this will drastically sooner or later affect
every Indic Wikisource projects and we have to plan accordingly.

Regards,
Bodhisattwa
Federico Leva (Nemo)
2018-03-15 20:11:39 UTC
Permalink
Just to eliminate a potential silly cause: do all those languages use
the same API key, and are you able to verify what's the quota usage?

Federico
Sam Wilson
2018-03-16 01:45:12 UTC
Permalink
They do all use the same API key, yes.

There doesn't seem to be anything obvious on the API end. Request numbers are pretty low under 1 request/minute and our quota is 600 r/m.

The tool seems to be working:

Loading Image...&lang=bn

Loading Image...&lang=

Maybe it's a problem with the gadget?
Post by Federico Leva (Nemo)
Just to eliminate a potential silly cause: do all those languages use
the same API key, and are you able to verify what's the quota usage?
Federico
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Sam Wilson
2018-03-16 01:46:19 UTC
Permalink
Sorry, I just realised you were all talking about the *Drive* API, not the *Vision* Api! :-) I dunno anything about that.
Post by Sam Wilson
They do all use the same API key, yes.
There doesn't seem to be anything obvious on the API end. Request
numbers are pretty low under 1 request/minute and our quota is 600 r/m.
https://tools.wmflabs.org/ws-google-ocr/index.php?image=https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fb%2Fbf%2FNandeesha_a.jpg%2F425px-Nandeesha_a.jpg&lang=bn
https://tools.wmflabs.org/ws-google-ocr/index.php?image=https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fa%2Fa2%2F07174-Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg%2F394px-07174-Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg&lang=
Maybe it's a problem with the gadget?
Post by Federico Leva (Nemo)
Just to eliminate a potential silly cause: do all those languages use
the same API key, and are you able to verify what's the quota usage?
Federico
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Jayanta Nath
2018-03-16 04:46:33 UTC
Permalink
Sam, the OCR works fine in Vision API in Bengali and other Wikisource ,
what you developed. Nothing worried. We will rewrite our python script for
running a Bot.
Post by Sam Wilson
Sorry, I just realised you were all talking about the *Drive* API, not the
*Vision* Api! :-) I dunno anything about that.
Post by Sam Wilson
They do all use the same API key, yes.
There doesn't seem to be anything obvious on the API end. Request
numbers are pretty low under 1 request/minute and our quota is 600 r/m.
https://tools.wmflabs.org/ws-google-ocr/index.php?image=
https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fb%2Fbf%
2FNandeesha_a.jpg%2F425px-Nandeesha_a.jpg&lang=bn
Post by Sam Wilson
https://tools.wmflabs.org/ws-google-ocr/index.php?image=
https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fa%2Fa2%
2F07174-Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-
Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg%2F394px-07174-
Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-
Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg&lang=
Post by Sam Wilson
Maybe it's a problem with the gadget?
Post by Federico Leva (Nemo)
Just to eliminate a potential silly cause: do all those languages use
the same API key, and are you able to verify what's the quota usage?
Federico
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Bodhisattwa Mandal
2018-03-20 03:12:10 UTC
Permalink
Update:

The OCR for Bengali using Google Drive API has stopped for pdf and djvu
files, but for jpg it is ok till now.

Shrini and Jayanta is rewriting and testing the OCR4Wikisource script to
convert the files into jpg files.

Looking forward,
Bodhisattwa

On 16 Mar 2018 10:16 am, "Jayanta Nath" <***@gmail.com> wrote:

Sam, the OCR works fine in Vision API in Bengali and other Wikisource ,
what you developed. Nothing worried. We will rewrite our python script for
running a Bot.
Post by Sam Wilson
Sorry, I just realised you were all talking about the *Drive* API, not the
*Vision* Api! :-) I dunno anything about that.
Post by Sam Wilson
They do all use the same API key, yes.
There doesn't seem to be anything obvious on the API end. Request
numbers are pretty low under 1 request/minute and our quota is 600 r/m.
https://tools.wmflabs.org/ws-google-ocr/index.php?image=https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fb%2Fbf%2FNandeesha_a.jpg%2F425px-Nandeesha_a.jpg&lang=bn
https://tools.wmflabs.org/ws-google-ocr/index.php?image=https%3A%2F%2Fupload.wikimedia.org%2Fwikipedia%2Fcommons%2Fthumb%2Fa%2Fa2%2F07174-Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg%2F394px-07174-Mei%25C3%259Fen-1906-Baarmanns_Bierk%25C3%25BChler-Br%25C3%25BCck_%2526_Sohn_Kunstverlag.jpg&lang=
Post by Sam Wilson
Maybe it's a problem with the gadget?
Post by Federico Leva (Nemo)
Just to eliminate a potential silly cause: do all those languages use
the same API key, and are you able to verify what's the quota usage?
Federico
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
_______________________________________________
Wikisource-l mailing list
https://lists.wikimedia.org/mailman/listinfo/wikisource-l
Loading...