Imagine using cuneiform to batch process a document there one page is a blank page, clearly you do not want the cuneiform program to report a crash for the blank page.
There are alternatives for what exactly should happen. Two options which would make more sense compared to reporting an error code:
* exit with a normal exit code and no output
* exit with a normal exit code with the output of the specified format (i.e. empty textfile or a .hocr file with the standard meta tags but no textlines etc).
Personally I would prefer the latter as an empty page is also important to be able to scan correctly (i.e. that no noise character are outputted for example).
Assume it is a test case to be written for cuneiform, say that version (A) of cuneiform exits with an error code as is today, (B) cuneiform exits with a normal status code but no output is produced, (C) as option (B) but with output produced.
test case for (A)
* process images, check if there is a crash and assume that means the page is blank, compare with ground truth
test case for (B)
* process images, check if no output was produced for any file and assume that means an empty page and compare with ground truth
test case for (C)
* process images, compare output with ground truth
and so producing the output and to check against it wit.
why do you think it's a bug? or you just need more clear message to
display?
--
Empty pages gives: PUMA_XFinalrecognition failed. https://bugs.launchpad.net/bugs/445357
You received this bug notification because you are a direct subscriber
of the bug.
Status in Linux port of Cuneiform: New
Bug description:
If given an image (jpg, or png or any type) it will fail if the image is completely blank (i.e all white).
$ cuneiform test.png
Cuneiform for Linux 0.8.0
PUMA_XFinalrecognition failed.
Imagine using cuneiform to batch process a document there one page is a blank page, clearly you do not want the cuneiform program to report a crash for the blank page.
There are alternatives for what exactly should happen. Two options which would make more sense compared to reporting an error code:
* exit with a normal exit code and no output
* exit with a normal exit code with the output of the specified format (i.e. empty textfile or a .hocr file with the standard meta tags but no textlines etc).
Personally I would prefer the latter as an empty page is also important to be able to scan correctly (i.e. that no noise character are outputted for example).
Assume it is a test case to be written for cuneiform, say that version (A) of cuneiform exits with an error code as is today, (B) cuneiform exits with a normal status code but no output is produced, (C) as option (B) but with output produced.
test case for (A)
* process images, check if there is a crash and assume that means the page is blank, compare with ground truth
test case for (B)
* process images, check if no output was produced for any file and assume that means an empty page and compare with ground truth
test case for (C)
* process images, compare output with ground truth
and so producing the output and to check against it wit.
Best regards,
Julien
_______ _______ _______ _______ _______ _____ gnition failed.
Från: <email address hidden> [<email address hidden>] för Serj Poltavskiy [<email address hidden>]
Skickat: den 5 november 2009 01:05
Till: julien
Ämne: [Bug 445357] Re: Empty pages gives: PUMA_XFinalreco
why do you think it's a bug? or you just need more clear message to
display?
-- gnition failed. /bugs.launchpad .net/bugs/ 445357
Empty pages gives: PUMA_XFinalreco
https:/
You received this bug notification because you are a direct subscriber
of the bug.
Status in Linux port of Cuneiform: New
Bug description:
If given an image (jpg, or png or any type) it will fail if the image is completely blank (i.e all white).
$ cuneiform test.png gnition failed.
Cuneiform for Linux 0.8.0
PUMA_XFinalreco