OCR!

[benzhen]

im considering getting an ipad just to scan my chinese books and read them. im wondering if there could be a mode for "reading" books that are sequences of jpegs. i have some awesome stuff in this format and it would be great if there could be a "page turn" function rather than having to exit out and select the next jpeg.

also i was wondering how the "scroll lookup words" function works on ipad. with so much screen, it seems unnecessary to have to scroll and zoom in order to recognize the text. it would be awesome if the green box could be dragged around, or just put in place by a tap.

also there is this book scanner being released, supposedly under the $200 mark. looks pretty amazing! http://www.ionaudio.com/booksaver

btw: i find that scanning the whole page is not as accurate as scroll lookup (especially when there is zhuyin or pinyin on the page). also i like leaving most of the page unadulterated for reading.

 

[mikelove]

Seems the iPad 2 has no auto-focus cam. I wonder how that will work with Pleco then. It works somehow on my iPod, but it's way better on an iPhone (with auto-focus - and the light also helps a lot).

Won't work very well, unfortunately - hopefully by the time the next iPod/iPad roll around somebody will come out with an autofocus camera that's thin enough for Apple to put in their non-iPhone devices.

im considering getting an ipad just to scan my chinese books and read them. im wondering if there could be a mode for "reading" books that are sequences of jpegs. i have some awesome stuff in this format and it would be great if there could be a "page turn" function rather than having to exit out and select the next jpeg

Certainly possible, but how would we know which one is the next JPEG? Would we just use straight ASCII order or is there some sort of common format for cataloging lists of page images (along the lines of EPUB)?

also i was wondering how the "scroll lookup words" function works on ipad. with so much screen, it seems unnecessary to have to scroll and zoom in order to recognize the text. it would be awesome if the green box could be dragged around, or just put in place by a tap.

We actually experimented a bit with that but found that we liked the scrolling interface better - you can't really position the box accurately when you're zoomed out to display the entire page at once, so you're going to have to drag around the background a lot anyway and at that point it's a lot easier if you don't have to drag anything else.

also there is this book scanner being released, supposedly under the $200 mark. looks pretty amazing! http://www.ionaudio.com/booksaver

Doesn't look like it supports automatic page turning, so I'm not sure how much of an advantage it'll really offer over existing tripod-plus-software solutions; certainly interested to see the reviews, though.

btw: i find that scanning the whole page is not as accurate as scroll lookup (especially when there is zhuyin or pinyin on the page). also i like leaving most of the page unadulterated for reading.

Yeah, unfortunately the full-page mode does tend to be easily confused by complicated layouts, which is one of the main reasons why we offer both. Pretty impressive that full-page OCR works at all for Chinese (and on a mobile device to boot), but perfect full-page capture is probably still years away...

 

[benzhen]

Certainly possible, but how would we know which one is the next JPEG? Would we just use straight ASCII order or is there some sort of common format for cataloging lists of page images (along the lines of EPUB)?

just order by image name in a folder?

We actually experimented a bit with that but found that we liked the scrolling interface better - you can't really position the box accurately when you're zoomed out to display the entire page at once, so you're going to have to drag around the background a lot anyway and at that point it's a lot easier if you don't have to drag anything else.

even with that much screen? hm i'll have to try it myself. i guess having half the page is better than what i have on my ipod right now.

Doesn't look like it supports automatic page turning, so I'm not sure how much of an advantage it'll really offer over existing tripod-plus-software solutions; certainly interested to see the reviews, though.

the fact the it has the cameras built in, is under $200, and saves directly to SD, and is all one piece, makes it more practical than anything else i've seen.

 

[mikelove]

just order by image name in a folder?

Straightforward enough, I guess, though numbered name ordering can actually be rather tricky (do we order by pure ASCII, do we try to parse numbers and order by them so that 9 would come up before 10, etc).

even with that much screen? hm i'll have to try it myself. i guess having half the page is better than what i have on my ipod right now.

Yeah, you can just about make it work if you view half the page in landscape but it still doesn't really feel nicer than scrolling the whole image.

the fact the it has the cameras built in, is under $200, and saves directly to SD, and is all one piece, makes it more practical than anything else i've seen.

The all-in-one aspect would be nice, yes, assuming they've integrated it well.

 

[benzhen]

Straightforward enough, I guess, though numbered name ordering can actually be rather tricky (do we order by pure ASCII, do we try to parse numbers and order by them so that 9 would come up before 10, etc).

im pretty sure all digital camera use 01,02...etc type numbering so that shouldnt be a big issue. maybe the group of users wanting to read entire scanned books is small but i thing it would be AMAZING. just my two cents....

 

[cjgait]

I temporarily have two iPads, my old iPad 1 that I haven't wiped yet to send to a relative and a new iPad2 that came in today. I first confirmed that the crappy camera on the iPad2 is not up to paper scanning due to low resolution and lack of autofocus, but another experiment worked, at least in demo mode. One of the most annoying things about Chinese books in the app store is how many of them don't allow cut and paste. For a native reader that is no big thing, but of course that is how I read: copy, then open in Pleco. So I tried aiming one iPad at the other on one of those books...and it works in demo mode! I was able to recognize characters and phrases on the iPad 1 from the iPad 2. I don't think I'm likely to go buy yet another iPad now, because I want to scan from paper books as well, but it is pretty cool. Next experiment: OCR on little tiny Chinese texts on my iPod Touch.

Regards,
Chris Gait
(Who is waiting for that classical dictionary this Summer and still, once in a while, would like to see 4-corner indexing available for when the HWR is stubborn).

 

[mikelove]

So I tried aiming one iPad at the other on one of those books...and it works in demo mode! I was able to recognize characters and phrases on the iPad 1 from the iPad 2. I don't think I'm likely to go buy yet another iPad now, because I want to scan from paper books as well, but it is pretty cool. Next experiment: OCR on little tiny Chinese texts on my iPod Touch.

Cool! A single-device alternative for that would be to take a screenshot from the ebook (press the power and home buttons together), then load that up in still-image OCR mode - should work pretty well assuming the ebook doesn't have an especially weird background pattern.

 

[cjgait]

Brilliant! I wish I'd thought of that. It works perfectly.