Hello, does anyone know of any such program that takes image files or pdf as input and works in different languages?
Quality at level of Evermote would be more than sufficient (if only Evernote would return the recognized text...).
I am happy with VisionObjects and am using it with different languages, but it does not work on image files.
Mac software for handwriting recognition?
10 posts
Page 1 of 1
-
AmberV
- Posts: 24793
- Joined: Sun Jun 18, 2006 4:30 am
- Platform: Mac + Linux
- Location: Ourense, Galiza
- Contact:
Multi-lingual OCR, basically? I wouldn't think that language would be necessary or useful for OCR, as this is just the recognition of characters and the translation of them from pictures into symbols. How those characters are strung together into words and in what language should be completely irrelevant to the pairing of a round shape with an 'o'.
.:.
Ioa Petra'ka
“Whole sight, or all the rest is desolation.” —John Fowles
Ioa Petra'ka
“Whole sight, or all the rest is desolation.” —John Fowles
AmberV wrote:Multi-lingual OCR, basically? I wouldn't think that language would be necessary or useful for OCR, as this is just the recognition of characters and the translation of them from pictures into symbols. How those characters are strung together into words and in what language should be completely irrelevant to the pairing of a round shape with an 'o'.
Amber, I am not familiar with handwriting recognition software but the problem seems not so different from OCR (optical character recognition). There, language is important because some inevitable ambiguities during OCR are usually weighted according to what is the more probable combination with the letter before and after and this depends strongly on the language used (e.g. the combination üen is unusual or absent in both English and German but does occur in Spanish). Hence most OCR programs let you indicate which language your source text was written in.
In my example, the OCR might be happy to accept the combination üen if it believed to be OCRing a Spanish text but be more inclined to evaluate alternative solutions if instructed to be looking at a text in German or English.
As I said I am not familiar with OCRing handwritten text so I cannot comment on this particular issue. But it is my understanding that OCRing printed text works precisely like that, i.e. language-aware, in Acrobat (the full version) and ABBYY. I haven't done any systematic tests to verify this because the results are good enough for me.
I thought Vision Objects software does work with image files, in the sense that at least some versions will OCR 'image' pdfs. To judge from one or two recent comments in the DevonThink forum, ABBYY, which DevonThink Pro and Pro Office use, will also OCR images of handwriting, but not very well; it will pick up keywords and titles if carefully written. But I imagine language is important, for reasons of context.
'Listen, some quiet night, when you've shirked your work that day. Do you hear
that distant, almost inaudible clicking sound? That's one of your
competitors, working away in the night in
Paris or London or Erie, PA.'
that distant, almost inaudible clicking sound? That's one of your
competitors, working away in the night in
Paris or London or Erie, PA.'
I bought ReadIris 11 some years ago. I’m a long time DEVONthink Pro user and did not upgrade for DT Pro Office, because I wanted a software capable to OCR handwritten texts (mostly family letters and some very old documents). That time DT had licensed ReadIris for it’s Office version, a kind of light version. After some discussions at the DT forums, I decided for the full ReadIris version, especially for my family documents I needed some of the advanced preferences.
OCR handwritten documents is possible, but it’s a pain. You may train the software and depending on the individual handwriting this can be a very, very long process. Don’t forget: to teach the software, you have to decipher the characters. If you are able to do so, you don’t need any more these OCR functions. I found that learning to read the old letters and documents is easier and more interesting, even if you have to write the text yourself. You will have less problems, if your handwritten documents are written by the same person and are written in ‚modern‘ style.
I did not upgrade to ReadIris 12 until now, I doubt there may be any improvement on deciphering handwritings. For machine written letters and documents the older version still works for me. (My documents are mostly written in German, yes, language is important.)
I just had a look at their website: http://www.irislink.com/c2-1685-189/Readiris-12-for-Mac.aspx. It seems they don’t advertise any more the ability to OCR handwritten documents. There’s only a hint regarding
„Handprinted notes recognition
Use I.R.I.S.’s unique handprinted notes recognition feature to transform your notes into editable text.“
I don’t know what this exactly means, maybe that's exactly what you are looking for. You should try the demo and see yourself. (You have to register for the demo and they will send you tons of Emails, don’t forget to unsubscribe if you don’t want them.)
OCR handwritten documents is possible, but it’s a pain. You may train the software and depending on the individual handwriting this can be a very, very long process. Don’t forget: to teach the software, you have to decipher the characters. If you are able to do so, you don’t need any more these OCR functions. I found that learning to read the old letters and documents is easier and more interesting, even if you have to write the text yourself. You will have less problems, if your handwritten documents are written by the same person and are written in ‚modern‘ style.
I did not upgrade to ReadIris 12 until now, I doubt there may be any improvement on deciphering handwritings. For machine written letters and documents the older version still works for me. (My documents are mostly written in German, yes, language is important.)
I just had a look at their website: http://www.irislink.com/c2-1685-189/Readiris-12-for-Mac.aspx. It seems they don’t advertise any more the ability to OCR handwritten documents. There’s only a hint regarding
„Handprinted notes recognition
Use I.R.I.S.’s unique handprinted notes recognition feature to transform your notes into editable text.“
I don’t know what this exactly means, maybe that's exactly what you are looking for. You should try the demo and see yourself. (You have to register for the demo and they will send you tons of Emails, don’t forget to unsubscribe if you don’t want them.)
@Ursula thanks for the encouragement, I will try ReadIris, I found their presentation of the mac software so unconvincing that I gave up before actually testing.
My application would be just transcribing my own notes, I am using Norwegian, German, and English.
I do have DT and my experience with ABBYY is that I cannot use it except for printed english text on scanned pdfs, but there are better programs out there for that. With my own writing, it did not work at all. Evernote is so much better, but for whatever reason they decided to just use recognition for search, not for output of text, I think this is very hard to understand.
VisionObjects does not do images/pdf and they actually make a major point out of that they not even intend to support it, which is sad because it works like magic. It reads my own handwriting better than I do.
I had hopes that there was something else out there but now it seems Readiris is the only option, I will report back once I tried it.
My application would be just transcribing my own notes, I am using Norwegian, German, and English.
I do have DT and my experience with ABBYY is that I cannot use it except for printed english text on scanned pdfs, but there are better programs out there for that. With my own writing, it did not work at all. Evernote is so much better, but for whatever reason they decided to just use recognition for search, not for output of text, I think this is very hard to understand.
VisionObjects does not do images/pdf and they actually make a major point out of that they not even intend to support it, which is sad because it works like magic. It reads my own handwriting better than I do.
I had hopes that there was something else out there but now it seems Readiris is the only option, I will report back once I tried it.
Have you considered reading the handwriting images aloud and using the Mac or Windows versions of Dragon Dictate to provide transcription? Not a cheap solution, but Vision Objects' software notwithstanding, my impression is that voice recognition is currently a more advanced and accurate technology than handwriting recognition - although both have some way to go.
'Listen, some quiet night, when you've shirked your work that day. Do you hear
that distant, almost inaudible clicking sound? That's one of your
competitors, working away in the night in
Paris or London or Erie, PA.'
that distant, almost inaudible clicking sound? That's one of your
competitors, working away in the night in
Paris or London or Erie, PA.'
Page 1 of 1