OCR!

[mikelove]

ETA: Prizmo needs fairly bright light to work, but it does work on the plain English text I tried it on. Interesting UI with lots of ways of improving the captured image (doesn't work 'live') for better OCRing.

Looks like more of a document management app with OCR only being one minor part - the image manipulation tools are interesting, but even in perfect conditions Chinese OCR from entire documents is a very dicey business, so I don't think we're likely to invest the time in developing tools like that for our experimental still image capture feature anytime soon.

 

[character]

Chinese class starts back up tomorrow -- any chance of a beta drop this evening? :wink:

OCR demos will sell copies of Pleco like no other feature ever did, I suspect.

 

[mikelove]

Chinese class starts back up tomorrow -- any chance of a beta drop this evening?

Sadly no, still going to be at least a few more days - we finally managed to come up with a way to auto-detect black-on-white versus white-on-black that's fast / reliable enough to be usable, but the UI needs some tweaking to accommodate that (an indicator for which mode it's in plus a way to toggle it / lock it into a particular mode if it gets it wrong), and there are a couple of unrelated 2.2 changes that have the potential to cause lots of bugs that we want to make sure are fully implemented for this release. (hardly anybody is willing to actually download / send us feedback on betas when we're not adding a big cool new feature to them, so we have to take advantage of opportunities like this)

 

[estudiando]

I like the idea of a standalone OCR application. I have a touch and "grand-fathered" Pleco....

 

[dustpuppy]

I also need pleco OCR for chinese classes
for on-computer chinese characters, I use MDBG, which is really awesome.

 

[character]

Long weekend coming up; perfect opportunity for beta testers to spend some quality time finding issues with OCR... :wink:

 

[mikelove]

Indeed it would be - FWIW, though, the time since the video hasn't been wasted, we've been adding a couple of very useful things including a flashcard scanning mode (point to a word for a second or so and it automatically adds it to flashcards), an alternate color for characters in the overlay to show you which ones are actually being used in the dictionary lookup (so you don't have to look down to see if it's "locked on" to the correct word), automated light/dark background detection, a (rather slow) option to use Sauvola binarization to improve accuracy when looking at pages that have lots of shadows / complicated backgrounds / etc... still adding a few more finishing touches, though.

 

[character]

Hey, I'll take it whenever. Just so it doesn't turn out to be 霧件. :wink:

Sauvola binarization

Nice idea: http://www.leptonica.com/binarization.html I never thought I'd have to invest in bright lights and a stand for my phone. Or my Chinese dictionary.

 

[mikelove]

Nice idea: http://www.leptonica.com/binarization.html I never thought I'd have to invest in bright lights and a stand for my phone. Or my Chinese dictionary.

Well it does slow things down quite a bit, so it'll be off by default and only recommended when you don't have a ready source of light handy (or find that the image is unreadable even with plentiful light), but it should be helpful in those specialized cases at least.

 

[CiiD]

Nice idea: http://www.leptonica.com/binarization.html I never thought I'd have to invest in bright lights and a stand for my phone. Or my Chinese dictionary.

Well it does slow things down quite a bit, so it'll be off by default and only recommended when you don't have a ready source of light handy (or find that the image is unreadable even with plentiful light), but it should be helpful in those specialized cases at least.

Speaking of which, if you haven't already, enabling usage of the LED on the iPhone 4/(hopefully the new iPod touch) for a possible light source may be useful. Only problem I could see is that the interface may become too convoluted with so many (useful) options.

 

[dustpuppy]

that's a bad idea, apple will ban your app.

 

[Entropy]

And the new iPod Touch will have both an HD video camera and a FaceTime camera.

Rejoice!

 

[mikelove]

Using the LED to illuminate camera capture is fine, they just don't like apps using it without the camera as a flashlight; we do indeed support LED illumination for live OCR, it's triggered by the lightning bolt icon that you can see in the video.

As for the new iPod, it's not yet clear whether or not it supports autofocus, but there's reason to suspect it doesn't. If it doesn't support autofocus, it can't do OCR - the resolution won't matter since it simply won't be able to take clear images at close range. Even the lowly 3G has enough resolution for live OCR, it can't do it because it lacks autofocus (and its processor is a bit too slow to do it well anyway).

So I'd hold off on ordering a new iPod for now, at least if you want OCR - you might be better off getting a used 3GS and not using it as a phone.

 

[mikelove]

That being said, the new iPod looks like a fantastic Pleco system in every other respect, so if you can live without OCR I'd certainly recommend it for people who don't want to get an iPhone. With our $70 educational-discount Pro bundle the 8GB model should make for a pretty damn amazing electronic dictionary for $300.

 

[caesartg]

Wow, that's really really disappointing if it turns out to be the case as the OCR had tipped it for me to get one of these rather than wait for Android. Do they use different camera modules for the 3GS/4 and the new Ipod Touch? Wonder if the limitation would be hardware or software. At any rate, once you know for sure, I definitely want to know. If it does turn out to have this limitation, I'm sure others who might be dashing out to buy the new iPod Touch model would also appreciate the heads-up.

 

[mikelove]

Wow, that's really really disappointing if it turns out to be the case as the OCR had tipped it for me to get one of these rather than wait for Android. Do they use different camera modules for the 3GS/4 and the new Ipod Touch? Wonder if the limitation would be hardware or software. At any rate, once you know for sure, I definitely want to know. If it does turn out to have this limitation, I'm sure others who might be dashing out to buy the new iPod Touch model would also appreciate the heads-up.

It's not the same module, no - it looks more similar to the module from the 3G. I haven't seen anyone explicitly say that it doesn't support autofocus yet, but the iPhone 4's specs page mentions autofocus while the iPod's doesn't, which is rather telling. Plus, the iPod has a very thin case, making it difficult to cram in an autofocus-capable camera, and since the camera is mainly designed to capture video rather than photos there's much less of a need for them to include autofocus than on the still-image-taking iPhone.

I suppose we could contact the folks behind RedLaser (which recognized barcodes even with non-autofocus cameras) and see if they'd be interested in licensing their image interpolation technology to use for use in OCR, but even if we did I'm not sure how accurate the results would be, and it would certainly be a lot slower / less "live" feeling. We could also offer an OCR module anyway but warn people that it would only work with objects distant enough for the iPod to focus on them correctly (like signs), but I'm not sure how useful that would be.

I've put up messages on our Facebook / Twitter pages telling people to hold off on pre-ordering a new iPod if they're interested in OCR, and hopefully one of the gadget blogs (or Apple themselves) will confirm soon whether or not the camera supports autofocus. As I said, though, it should still be a lovely Pleco device if you don't mind the (probable) lack of OCR - Pleco on a Retina Display is gorgeous and the A4 processor is screaming fast.

 

[caesartg]

This article http://www.appleinsider.com/article...ith_retina_display_forward_facing_camera.html suggests that it's the same two cameras. Of course, they might be talking rot. There are a lot of companies here in Taiwan selling phone camera components and it's pretty competitive, so it's possible that the components are different and are limited in different ways. I notice how the new spec-sheets on the US store say that the 3GS is

Tap to focus video or still images

while the new Ipod Touch is

Tap to control exposure for video or stills

. Where's the darn quick answer 'chat' button mentioned in Apple's support pages? Grrrr, no way to ask the question without a North American phone number. Although somehow I imagine Apple's retail staff will be 10000% busier than usual just now fielding all sorts of questions.

Ah, I notice you've already responded on the above, cheers. Yes, you're right it will still be an amazing Pleco device, but the OCR tipped it for me (although, I have to say, Mr Jobs has done a fine job of convincing me that I need to buy all Apple's products, including the touchscreen Nano, the little TV box and to sign up for Ping®!) and I still have my trusty old Dell Axim (sigh :cry: ).

 

[radioman]

I don't want to get off on a rant here, but, as I sit here in Dalian throwing down a double espresso with my 3Gs, I cannot help but comment on what a lovely OCR concept Pleco brings to the table and the timing of the new iPod touch which delivers everything... but a decent camera. Just great.

Personally, I would have preferred the device to be a bit thicker with a bigger battery and better camera. But that's just me. I think with that camera, I likely will not be getting one.

 

[mikelove]

I don't want to get off on a rant here, but, as I sit here in Dalian throwing down a double espresso with my 3Gs, I cannot help but comment on what a lovely OCR concept Pleco brings to the table and the timing of the new iPod touch which delivers everything... but a decent camera. Just great.

Personally, I would have preferred the device to be a bit thicker with a bigger battery and better camera. But that's just me. I think with that camera, I likely will not be getting one.

It's well-designed for its purpose, I think - I would guess that the three most important groups of iPod Touch buyers are:

• Pre-teens / kids who don't have cell phones (seriously a huge market), who will be delighted with even a simple little camera to take pictures of their friends
• People who can't afford / don't want to be burdened with full-blown smartphones but want access to App Store (or just a really cool device for playing music / watching movies), who are going to value extreme portability since they'll likely be carrying a regular dumbphone too
• Android / BlackBerry users who want access to App Store / superior media support - again, this is being carried along with their regular cell phone so the thinner the better, and battery exhaustion isn't that big an issue since they'll still have their smartphone working fine

So it should be a big hit for Apple - might take over a large chunk of the market for Flip-like low-end camcorders, among other things. It's just not so great for the developers of autofocus-dependent OCR / barcode reader / etc apps

 

[radioman]

Designed for their purpose... Yeah they are VERY good about that. For this, I have no doubt. Was just doing a bit of venting, stating MY purpose/hope. But I guess it ain't a "custom build" for Radioman

But it might work great for my kids as they would be able to easily video conference me when I'm on the road. For that matter, I wonder when they will drop Facetime into iChat, etc. on the Mac. I guess that will be later as they don't want people thinking that buying one phone/device for a family is good enough, at least not yet.

I'm sure I'll go the iPhone 4 root at some point. I'm very happy with the 3Gs w/4.0. But having the new iPod Touch with a great camera - well, the timing with the OCR coming out would have been sweet.

And who knows, maybe the OCR it will work on some level. If there is a practical way to make it work, I'm sure Pleco Software Incorporated will figure it out.