snsok 0 Posted March 1, 2010 Posted March 1, 2010 I am still looking for a solution to keeping sets of handwritten notes in Evernote and making them searchable. I would like to scan them as a batch to a multi-page file.I am a Premium user.I would like to get handwriting recognition of these handwritten notes. I understand that PDF files will have text recognition but not actual handwriting recognition.I tried multipage TIFFs but as I read on the forum from a year or two ago, EN doesn't (or at least didn't) support multipage. Is that still the case in the context of handwriting recognition.Another option seems to be stacking images into a single note. I am able to add 2 or 3 TIF images to a single note. If I do that, will handwriting recognition function?Batch scanning a stack of notes from a single meeting or single class or even single day would be a very nice attribute.Thanks
engberg 89 Posted March 1, 2010 Posted March 1, 2010 For handwriting, you will get your best results with JPEG. Unfortunately, JPEG is a single-page format, so you'd need to create one file per page.
pdondl 0 Posted March 27, 2010 Posted March 27, 2010 I assumed that the very good OCR for handwritten text would be available now for PDFs, after image PDFs were made searchable in July last year - so I got a premium subscription.Why does the PDF-OCR only work (well) for printed text? It's a complete mystery to me, and many people seem to be interested in having some way of dealing with multi-page handwritten documents.I surely hope that there will be an option to use the jpeg OCR for PDFs soon - it can't really be technically difficult, since those PDFs are just multiple images (TIFF, JPG or PNG, I believe) in a PDF enclosure. Otherwise, that was quite a waste of money.BTW: The links to 'feedback' and 'suggestions' in the FAQ don't work.
BurgersNFries 2,407 Posted March 27, 2010 Posted March 27, 2010 Why does the PDF-OCR only work (well) for printed text?viewtopic.php?f=30&t=15182#p59616I assumed that the very good OCR for handwritten text would be available now for PDFs, after image PDFs were made searchable in July last year - so I got a premium subscription.I surely hope that there will be an option to use the jpeg OCR for PDFs soon - it can't really be technically difficult, since those PDFs are just multiple images (TIFF, JPG or PNG, I believe) in a PDF enclosure. Otherwise, that was quite a waste of money.I don't know why people tend to assume the things they want are not difficult to add. (shakes head) I would also "assume" if that's the primary reason for you to go premium, you would have done your due diligence & spent $5 for a monthly subscription to confirm it works & works the way you understand it does? So you wasted the price of a vente mocha at Starbucks? If, OTOH, you did drop the annual fee of $40, if you search the forum on posts by user "engberg" with the word refund, you'll see he's always happy to refund your subscription, if you're not happy.Also, if you read the link I posted, it may well be that applying the jpeg OCR to PDFs would increase the false positives when the PDF is all/mostly text.
pdondl 0 Posted March 28, 2010 Posted March 28, 2010 viewtopic.php?f=30&t=15182#p59616I saw that now, too -- after reading the PDF-OCR blog post I got excited, tried it out, and then tried to figure why it didn't recognize any of the text.I don't know why people tend to assume the things they want are not difficult to add. (shakes head)Because I program quite a bit myself, and so I know that a PDF in the case of a scanned note is just a container for one raster image (sometimes even JPEG encoded) per page. This makes it not much different from a note that is composed of multiple JPEGs, only that it can be produced with much less effort in a scanner. And your link shows that I'm not the only person who writes more than one page at a time by hand. Also, if you read the link I posted, it may well be that applying the jpeg OCR to PDFs would increase the false positives when the PDF is all/mostly text.Well, that sort of is the big advantage of Evernote - nothing I have seen comes close to its handwriting recognition. False positives are not a big problem for searching, not having a reasonable way to use multi-page documents is. It just seems like Evernote is only an inch away from being perfect for archiving handwritten notes, but the lack of support for more than one page makes it useless for many people (at least judging from the forums). It's such a great program, and without any alternatives - that's why I'm hoping that they'll make it usable for this purpose, too.
engberg 89 Posted March 28, 2010 Posted March 28, 2010 Our handwriting recognition for JPEG produces a hidden internal search "tree" for the image, which lists a cloud of possibilities for every word-shaped region. This representation isn't very useful for actually extracting the text, it's only good for searching against these clouds within Evernote. However, it's helpful for finding lower-quality images within Evernote.Our text processing for PDF is basically using best-of-breed OCR software to produce a second PDF document that we internally index. This PDF is more usable for people that want to (e.g.) copy and paste text out of a cleanly scanned document, but doesn't provide a "cloud" of possibilities for each word, which reduces the results for handwritten text noticeably.I.e. these solutions are tuned to be as useful as possible for two slightly different use cases.Our R&D team is making continual improvements on the image processing quality and results, and we believe we can do some improvements for handwriting in the future.
pdondl 0 Posted March 28, 2010 Posted March 28, 2010 I.e. these solutions are tuned to be as useful as possible for two slightly different use cases.I realize that Evernote's handwriting recognition would only be usable within the program itself - that's fine. My scenario is that I have a lot of workshop, meeting, etc. notes that I would like to organize and make searchable (within Evernote is fine) - only most of those notes are longer than a page, so it's preventatively cumbersome to import them into Evernote in a reasonable way (e.g., as notes including more than one JPG).Having tried with a few using JPG, I have noticed that Evernote works great with my handwriting, which is nothing short of amazing. The best thing I could think of would be some kind of option of whether to use the print-OCR or handwriting-OCR on a pdf. Maybe some time in the future, it did seem like you get this request often.
BurgersNFries 2,407 Posted March 30, 2010 Posted March 30, 2010 False positives are not a big problem for searching, Maybe not for you. But I suspect there are a lot of others who would prefer to not have to filter through all the false positives when dealing with clean text rather than images.
Recommended Posts
Archived
This topic is now archived and is closed to further replies.