really bad recognition in Windows version

[Marko ŽZmak]

I tried ocrad for windows (downloaded from: http://sourceforge.net/project/showfiles.php?group_id=61702&package_id=200509) but I get really bad results. For example I took a screenshoot of this page:

http://spambayes.sourceforge.net/download.html

cut out the first paragraph and saved it as ppm.

I tried ocrad on it but it recognized lees than every other character. Here is the output:

ve__lon _ O o o_ _he _DamBaUe_ D_olec_ |_ now a_allable
Thl_ |_ a b_g_lx _elea_e - |_ |_ __n_lonallu lden_lcal _o _ O, b__ lncl_de_ _lxe
_ _o_ a n_mbe_ o_ b_g_ We exDec_ |_ _o D_o_e _o
be o_|_e __able and __able bu mo__ DeoDle, a_ a _e__|_ (and _lnce we a_e ha_d a_
 wo_k on _he _Dcomlng _ _ _elea_e) we
exDec_ _ha_ _hl_ wlll be _he la__ _elea_e ln _he _ o x llne _eedback _o _
The _econd alDha _elea_e o_ _ _ |_ al_o now a_allable I_ |_ hlghlu llkelu _ha_ _
he_e a_e new b_g_ ln _hl_ _elea_e (e_DeclallU
wl_h _he IMAP _||_e_), b__ |_ uo_ a_e wllllng and able _o gl_e |_ a _Dln _o_ __,
 _ha_ wo_ld be g_ea_lu aDD_ecla_ed vo_ mlgh_
llke _o look a_ _hl_ _
vo_ mau llke _o _lew _he _elea_e no_e_ o_ _he _
Micro5oft Window5
Mlc_o_o__ Wlndow_ __e__ a_e enco__aged _o __e _he ln__alla_lon D_og_am _o ln__al
l _DamBaUe_ Thl_ wlll ln__all aDDllca_lon_
__|_able _o_ almo__ all emall cllen__, lncl_dlng Mlc_o_o__ O__look and Mlc_o_o__
 O__look _xD_e__
Plea_e _ee o___ o_l_mD dl_ec_lu _o _he _
Source Relea5e5
The _o__ce-code _elea_e_ can be __ed on anu Dla__o_m wl_h a pu_hon ln_e_D_e_e_
The _o__ce-code _elea_e_ can be downloaded, a_ el_he_ a g_IDDed _a_ball o_ a _ID
 _lle, dl_ec_lu __om _he _lle _elea_e_ _o_ _hl_
_o ec_
p_e_eo_|_|_e_


I tried changing some command line options (size, threeshold), but nothing helped.
Could someone help me out with this? Is ocrad really so bad at recognizing text, or do I have to change some options in order to recognize the text better?

[Leon Kolchinsky-2]

>
>
> I tried ocrad for windows (downloaded from:
> http://sourceforge.net/project/showfiles.php?group_id=61702&package_id=200
> 509)
> but I get really bad results. For example I took a screenshoot of this
> page:
>
> http://spambayes.sourceforge.net/download.html
>
> cut out the first paragraph and saved it as ppm.
>
> I tried ocrad on it but it recognized lees than every other character.
> Here
> is the output:
>
> ve__lon _ O o o_ _he _DamBaUe_ D_olec_ |_ now a_allable
> Thl_ |_ a b_g_lx _elea_e - |_ |_ __n_lonallu lden_lcal _o _ O, b__
> lncl_de_
> _lxe
> _ _o_ a n_mbe_ o_ b_g_ We exDec_ |_ _o D_o_e _o
> be o_|_e __able and __able bu mo__ DeoDle, a_ a _e__|_ (and _lnce we a_e
> ha_d a_
>  wo_k on _he _Dcomlng _ _ _elea_e) we
> exDec_ _ha_ _hl_ wlll be _he la__ _elea_e ln _he _ o x llne _eedback _o _
> The _econd alDha _elea_e o_ _ _ |_ al_o now a_allable I_ |_ hlghlu llkelu
> _ha_ _
> he_e a_e new b_g_ ln _hl_ _elea_e (e_DeclallU
> wl_h _he IMAP _||_e_), b__ |_ uo_ a_e wllllng and able _o gl_e |_ a _Dln
> _o_
> __,
>  _ha_ wo_ld be g_ea_lu aDD_ecla_ed vo_ mlgh_
> llke _o look a_ _hl_ _
> vo_ mau llke _o _lew _he _elea_e no_e_ o_ _he _
> Micro5oft Window5
> Mlc_o_o__ Wlndow_ __e__ a_e enco__aged _o __e _he ln__alla_lon D_og_am _o
> ln__al
> l _DamBaUe_ Thl_ wlll ln__all aDDllca_lon_
> __|_able _o_ almo__ all emall cllen__, lncl_dlng Mlc_o_o__ O__look and
> Mlc_o_o__
>  O__look _xD_e__
> Plea_e _ee o___ o_l_mD dl_ec_lu _o _he _
> Source Relea5e5
> The _o__ce-code _elea_e_ can be __ed on anu Dla__o_m wl_h a pu_hon
> ln_e_D_e_e_
> The _o__ce-code _elea_e_ can be downloaded, a_ el_he_ a g_IDDed _a_ball o_
> a
> _ID
>  _lle, dl_ec_lu __om _he _lle _elea_e_ _o_ _hl_
> _o ec_
> p_e_eo_|_|_e_
>
>
> I tried changing some command line options (size, threeshold), but nothing
> helped.
> Could someone help me out with this? Is ocrad really so bad at recognizing
> text, or do I have to change some options in order to recognize the text
> better?
> --

>From my experience I can say that ocrad recognition is great with FuzzyOCR.
I'm using ocrad-0.16 version.
What version are you using?

Try to use ocrad with the following options:
giftopnm image001.gif > image001.pnm
# ocrad -s5 image001.pnm
# ocrad -s5 -i pic image001.pnm


Best Regards,
Leon Kolchinsky






_______________________________________________
Bug-ocrad mailing list
Bug-ocrad@...
http://lists.gnu.org/mailman/listinfo/bug-ocrad

[Marko ŽZmak]

I'm using version 0.14. And yes, changing the scaling factor to 5 did help, thanks.

When I tried changing the scaling factor, I tried several values, but not 5. How did you guess what scaling factor should I use? Is there any "recipe" from which I can see what scaling facto should I use for my images?

Leon Kolchinsky-2 wrote:

From my experience I can say that ocrad recognition is great with FuzzyOCR.
I'm using ocrad-0.16 version.
What version are you using?

Try to use ocrad with the following options:
giftopnm image001.gif > image001.pnm
# ocrad -s5 image001.pnm
# ocrad -s5 -i pic image001.pnm


Best Regards,
Leon Kolchinsky

[Leon Kolchinsky-2]

>
> I'm using version 0.14. And yes, changing the scaling factor to 5 did
> help,
> thanks.
>
> When I tried changing the scaling factor, I tried several values, but not
> 5.
> How did you guess what scaling factor should I use? Is there any "recipe"
> from which I can see what scaling facto should I use for my images?

Just a member of FuzzyOCR mailinglist, there we share the best options for OCR software like gocr,ocrad, tesseract etc.

You can get some tips from FuzzyOCR 3.5.1 documentation :)

>
>
>
> Leon Kolchinsky-2 wrote:
> >
> > From my experience I can say that ocrad recognition is great with
> > FuzzyOCR.
> > I'm using ocrad-0.16 version.
> > What version are you using?
> >
> > Try to use ocrad with the following options:
> > giftopnm image001.gif > image001.pnm
> > # ocrad -s5 image001.pnm
> > # ocrad -s5 -i pic image001.pnm
> >
> >
> > Best Regards,
> > Leon Kolchinsky
> >
>
> --
> View this message in context: http://www.nabble.com/really-bad-
> recognition-in-Windows-version-tf3238133.html#a9037909
> Sent from the Gnu - Ocrad mailing list archive at Nabble.com.
>
>
>
> _______________________________________________
> Bug-ocrad mailing list
> Bug-ocrad@...
> http://lists.gnu.org/mailman/listinfo/bug-ocrad

_______________________________________________
Bug-ocrad mailing list
Bug-ocrad@...
http://lists.gnu.org/mailman/listinfo/bug-ocrad