Re: [orca-list] The new Hypra website



Actually user is free on our system. If a person knows an OCR tool which
covers his/her usage, anyone can install it (from repo, from anywhere).
If Tesseract covers needs, mate-accessibilitz-ocr has it.

But for those who have low performance, they need to know the
proprietary solution exists, even on a free software environment. It's
an optional solution if free solutions aren't enough. We do>t make any
tool mandatory for a feature, except the desktop environment.

Regards,


Le 27/08/2016 à 18:17, B. Henry a écrit :

Yes, I really did not want to write about this again, but as you must not have seen the discussion in the 
past...
The free engines work fine for many people, and fine for some or many jobs for others, but for another 
groups of us who do things that require near perfect 
results, i.e. to fill out paperwork that has life impacting importance, read mail that may be in much less 
clear fonts than those you have recognized on 
your computer screen with ocrdesktop, etc, yes, there is for sure a reason to use something that gives 
better results.
I can't compare abbyy to teseract as I've not used the former, but I can compare the results using lios or 
something else using teseract to the results I 
got using kurzweil, and the difference on more than one occasion, was between around 70% accuracy and 92% 
accuracy. That is major, and not only that, but 
70% accuracy is not good enough to do more than see what something is more or less, e.g., would you get 
anything out of a bank statement that was only 70% 
read correctly?
I do not get bank documents, or anything else in paper if I can get it on line as a pdf, but the whole 
point of the program, as far as I know lios is what 
is being worked on,
is to give folks something good enough to do as much as is possible with, i.e. not a something that won't 
work for many people in its target user base.
I do not use other opperating systems other than Linux on my PC(s) unless there is a document I can not 
OCR. I honestly can think of nothing else other 
than occasional curiosity that has gotten me to use my one installed copy of windows in years, but I must 
keep a working windows to run kurzweil, and I can 
imagine myself traveling, and traveling I'd not have my windows box with me ever, where I'd need to read a 
paper of some kind. 
I can not say to someone who has always gotten very good results using  cuniform or teseract that that is 
not there experience, but I've seen the 
difference in results between  the better proprietary options and the options from the FOS world. 
Also, as long as I am writing this, the one major difference between the "eloquence case" as another poster 
called it and this is that the old ibm-tts has 
not been developed in years. Abbyy is a maintained piece of software, and there are other proprietary 
engines that are maintained and or under development 
that might become available, or be made available for us on Linux. 
Would I prefer seeing tesseract improve to at least with in striking range of the accuracy acheived by the 
best solutions available for personal computers, 
hell yes! If that isnot possible would I like to see some one write a new rockin FOS OCR solution, again 
for sure and of course.
I do not know of such work, but as I said before, then when it becomes available we can use the free 
engine(s), but until then let's not try and tell 
people they can't work, and or must get a lot of extra sighted help. 
If you do not need the proprietary engine, then of course you are more than free not to use it, but do not 
try and hold those who do need it back.
I'm done, said this stuff before, tried to be clear and maybe even persuasive, but hey, the fsf has not 
even put a priority on accessiblity that I've ever 
seen. 
What, one of their darling distros is nicely accessibler for blind folks, another completely unusable from 
what I hear, and...can't remember about the 
other for sure.
My point being, we are not necesarily going to get any help from the hard core free software movement, nor 
have I seen any new free OCR projects mentioned 
in any context, so it's posibly not just a case of holding out for a few months. 
I get the propaganda because I strongly support FOS, even started a google group to promote the discussion 
of accessible free software, but because some 
proprietary projects exist in the world, and maybe some of us need to use a few of them, that does not make 
free software less valid, not as good or 
anything else other than what it is and can be.
Actually there's a case to be made that a good proprietary project can put pressure on the good free 
project to become better, not just a "poorman's 
solution".
I apologize for the verbose conclusion, bjut really want to move on to other things and wanted to make the 
major points and be done with it.
Regards,  
 
    
   


-- 
Jean-Philippe MENGUAL

HYPRA, progressons ensemble

Tél.: 01 84 73 06 61

Mail: contact hypra fr

Site Web: http://hypra.fr


[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]