[bksvol-discuss] Re: txt page breaks redux
- From: "Sarah Van Oosterwijck" <curiousentity@xxxxxxxxxxxxx>
- To: <bksvol-discuss@xxxxxxxxxxxxx>
- Date: Wed, 29 Dec 2004 11:52:42 -0600
Mine isn't either. It's a 1Ghz with 512 MB of slow RAM. I can see the
difference scanning in grey scale causes. I have about 320 characters being
recognized per second with static threshholding and 300 DPI. I can't scan
at 400 DPI, but sometime I will scan in grey scale to see the difference. I
wouldn't count the book I did scan in grey scale as something normal to use
for comparison because it had to use multiple languages and the text was
pretty awful.
My scanner is slower, so I prefer to edit a little over rescanning pages
that aren't too bad, or even scanning using slower settings. I'm sure there
are others that feel the same about that. Besides, reading and fixing is a
good deal more stimulating to the brain than holding down a book while it
scans. ;-)
Sarah Van Oosterwijck
http://home.earthlink.net/~netentity/
----- Original Message -----
From: "Guido Corona" <guidoc@xxxxxxxxxx>
To: <bksvol-discuss@xxxxxxxxxxxxx>
Sent: Wednesday, December 29, 2004 9:45 AM
Subject: [bksvol-discuss] Re: txt page breaks redux
> Sara, in the last book reco speed was approx 170 chars per second. That
> would have meant approx 20 seconds of reco time per double page. Scanning
> takes about 7 seconds per double page, plus return time.
> My machine has a mobile Pentium M running at 1.6Ghz with 1.5 GB of RAM.
> This is roughly equivalent to a 1.1Ghz Pentium 4 on a desktop.
> Not the fastest kid on the block.
>
> Guido Dante Corona
> IBM Accessibility Center, Austin Tx.
> Research Division,
> Phone: 512. 838. 9735.
> Email: guidoc@xxxxxxxxxxx
> Web: http://www.ibm.com/able
>
>
>
>
> "Sarah Van Oosterwijck" <curiousentity@xxxxxxxxxxxxx>
> Sent by: bksvol-discuss-bounce@xxxxxxxxxxxxx
> 12/28/2004 11:26 PM
> Please respond to
> bksvol-discuss
>
>
> To
> <bksvol-discuss@xxxxxxxxxxxxx>
> cc
>
> Subject
> [bksvol-discuss] Re: txt page breaks redux
>
>
>
>
>
>
> Just for my curiosity, how fast is your computer, and how many seconds
> does
> each scan take? Have you ever reset your recognition statistics when you
> started a new book and did the math to figure out exactly how fast the
> average character, or paperback page takes to recognize. You seem to be
> interested enough in statistics to have tried it.
>
> Online you can find all kinds of benchmarks for processors, but the only
> kind I would really be interested in would be an OCR benchmark. hehehe
>
> BTW, I haven't done the math on the recognition statistics for my
> computer,
> because I don't like math enough to do it just for the fun of it, and i
> had
> no other computer to compare with.
>
>
> Sarah Van Oosterwijck
> http://home.earthlink.net/~netentity/
>
> ----- Original Message -----
> From: "Guido Corona" <guidoc@xxxxxxxxxx>
> To: <bksvol-discuss@xxxxxxxxxxxxx>
> Sent: Tuesday, December 28, 2004 9:12 PM
> Subject: [bksvol-discuss] Re: txt page breaks redux
>
>
> > Kelly, if the scanner is the cause of the quality problem, a minimal
> $45
> > investment will get them a perfectly good EPSON 1650 at the EPSON refurb
> > store. If someone can't afford that, I can't even see how they can't
> > possibly afford a monthly Internet connection charge.
> > A barely higher $124 will get them a modern refurb EPSON 3170 in the
> same
> > place. If Kurzweil or Openbook were too costly and rehab funding were
> not
> > available, the ABBY Fine Reader Professional 7.0 is a perfectly high
> > quality solution, as the spottless submission from Donna Smith testify.
> > As you said, obsolete equipment is not a good excuse. A little
> up-front
> > work prior to submission typically ensures that a good part of the
> errors
> > have been fixed. A spotcheck can also detect bunch of missing words
> etc.
> > . .
> >
> > Now, think about time usage: is it better to work 20 hours to salvage
> a
> > single book, or spend the same total amount of time and end with 3 to 4
> > submissions at the end? Let's not even think about the fact that our
> > paying subscribers will find the collection grown by 4 instead of 1 at
> the
> > end of your effort. Let us think about our work benefitting other
> > volunteers? Your 20 hours can be spent giving credit to 1 sloppy
> > submitter, or give credit to 4 other good ones. Now, tell me where
> you
> > will work, if the greater good of the volunteer community is paramount
> to
> > you.
> > And if instead you think about your own credits, as a reviewer, your 20
> > hours can get you 1 credit, or can get you 4, depending on what you
> work
> > on.
> >
> > So, as you can see, whether you think about our customers, the
> volunteer
> > community at large, or your own interest, the outcome seems to be the
> > same.
> >
> > Unless we think of these files as orphan, hungry, sick children, in
> need
> > of comfort, and nurture. Which I am afraid they are not.
> >
> > Guido
> >
> >
> >
> >
> >
> >
> >
> > Guido Dante Corona
> > IBM Accessibility Center, Austin Tx.
> > Research Division,
> > Phone: 512. 838. 9735.
> > Email: guidoc@xxxxxxxxxxx
> > Web: http://www.ibm.com/able
> >
> >
> >
> >
> > "Kellie Hartmann" <kellhart@xxxxxxxxxx>
> > Sent by: bksvol-discuss-bounce@xxxxxxxxxxxxx
> > 12/28/2004 08:19 PM
> > Please respond to
> > bksvol-discuss
> >
> >
> > To
> > <bksvol-discuss@xxxxxxxxxxxxx>
> > cc
> >
> > Subject
> > [bksvol-discuss] Re: txt page breaks redux
> >
> >
> >
> >
> >
> >
> > Oh Guido, great Lord High Rejecter of all, <grin>
> > That's fantastic if you have access to a copy of the book. Oftentimes
> the
> > volunteer may not, and a certain amount of error-correcting really isn't
> > that onerous, especially if you're going to read the book anyway.
> > I do think, though, that it's nice when people who, because of older
> > equipment, can't get scans that live up to our modern high standards go
> > through and do some work on their submissions before submitting. After
> > all,
> > there's a lot more credit for submitting than there is for validating.
> > Kellie
> >
> >
> >
> >
>
>
>
>
Other related posts:
