Slashdot Mirror
Word Lens — Augmented Reality Translation
Barence writes "PC Pro has a review of a new augmented reality iPhone app that translates from Spanish to English on the fly. 'Point the camera at a decent-sized chunk of Spanish text and within a couple of seconds you'll get a rough and ready translation,' said the reviewer. 'And most magnificently of all, the translation is overlaid, at the correct size, on the original object.' The team behind the project has produced a video of Word Lens in action."
This is pretty damn cool. But no android app. No news if they plan on releasing one. In fact, their site is pretty void of any information at all. I would buy this just to play with it, but I'll never be an iphone guy.
Belief? Hope? Preference?The Existential Vortex
I expect mistakes of Newton hand writing recognition like proportions. Let the laughs begin.
Okay let's see, we combine the terror of OCR with mangled language translation and the pit fall of cropped or intersecting text patches and variable fonts and multiple contexts? My hovercraft is indeed full of eels.
Some drink at the fountain of knowledge. Others just gargle.
We *really* need some practical computer goggles/glasses already. Think about how much more badass this would be if you didn't even have to hold a phone in front of your face.
I think I could life with low framerate in the real world if it got me stuff like this.
Someone was showing this off at TechShop last evening. Very nice.
i remember the days when all the new cool tech was only seen in the government and large corporations first and then trickled down to us peons. these days with our rampant consumerism it's the opposite. we see cool stuff like this first and it's cheap and the big boys are now playing catch up because things move so fast
if it wasn't for our vane consumerism this would be a government project costing tens of millions of $$$ in R&D and the devices would be single use devices that also cost some ridiculous amount of money
.... I'll never be an iphone guy.
It's hard to get the meaning over the internet. Here's an example...
First guy sees another guy that attracts him. "So, are you an iPhone guy?", he asks with a sly grin.
Second guy, "Why yes I am. And I have a Hummer. Want a ride?"
"Sure!"
That's how a lot of homsexual pickups go these days. Or you'll see gay guys drive around in Hummers with their iPhones prominently displayed. The Hummer is a physical display that you give "hummers", i.e. blowjobs. The possession of the iPhone is sefl explaining.
To just teach the Spanish to speak English? Why do I have to make accommodations because of their choice to speak gibberish?
Too bad that this system only works with limited amount of texts. I installed this app on my iPod Touch and tried the default text reversing filter. If I used a Serif font, this could not read the words realiably. Fonts needed to be Sans Serif. Also this uses some dictionaries so if the word is not in dictionary (eg. deemed offensive) or some random gibberish, this could not recognize it. And all this I did with large black text on white background so viewing conditions are definitely not the issue.
This looks cool. now, if I hold it up to my ear will it translate speech in real-time too?
Well, we just need a proper translation. "Y lo va el otro direccion"?? Please...
Now they just need to do the same for economically relevant languages. The top developing countries currently are Brazil, India, and China (in no particular order) and none of them speak Spanish as a primary language.
Of course, I tried to use a similar argument decades ago in school when everyone told me I needed to learn Spanish (while living in a state that was dramatically closer to French Canada than to Mexico, but oh well), and I still ended up taking three years of a language that I almost never encounter in my regular existence.
Mandarin Chinese, on the other hand, I hear every day at work. In my work Espanol is marginally more valuable than Esperanto. But what do I know, really...
Damn_registrars has no butt-hole. Damn_registrars has no use for a butt-hole.
I'd like to see the option to translate from English to Mandarin Chinese behind the scenes, then display a translation from Mandarin right back to English. Augmented Engrish!
Actually, I could use something like that... I live in the Sunset Park neighborhood in Brooklyn and for a few blocks in every direction nearly everything is in Spanish. Most of the shops have names in Spanish, packages inside are often Spanish... And I don't read Spanish...
I remember back in the early 90s when a guy showed me an 8Gb backup tape he had in his shirt pocket and I thought, Holy Crap, 8 GIGA-bytes fits in your pocket now? That's Awesome! And now, years later, you can carry many times that much data on a keychain. Equally Awesome. And this, this translator thing... totally and completely Awesome and Amazing. If you picture yourself as someone from say 100 years ago looking at today's world, some things we take for granted are pretty much like magic.
The translation from English to Spanish is pretty shitty.
I rarely respond to comments. Also, don't ask for clarifications: a brain and Google are faster, believe me!
These guys just opened a gold mine.
I'm sure there will be a ton of cynical and jaded comments here, but this is a working prototype of augmented reality that is actually immediately and unquestionably useful, even in its infant state. Even non-technical people can see the promise of this, and graspable promise equals investment.
Bravo, and congratulations to the developers!
ATM, this post is modded zero-Troll. For reals? Has everyone lost their sense of humor? Wow.
Like the OP really thinks that everyone in the world should be taught to speak English instead of the "gibberish" they speak now??
Maybe it has something to do with geeks leaning toward autism and autistics taking things too literally?
Just read this on MSNBC. The author shows what happens when trying it on basic Spanish.
Overall, not worth the money until it gets heavily reworked.
We will bankrupt ourselves in the vain search for absolute security. -- Dwight D. Eisenhower
... when he says that one of the reasons the iPhone won't run Flash is because it doesn't have the processing power.
You're telling me it can have the power needed to do something like this - analyze an image for text, decode it, put sentences together, translate, match the most appropriate font and colours, scrub the original text, render the new text at the appropriate angle and position - but not to play Flash movies. I call bullshit.
== Jez ==
Do you miss Firefox? Try Pale Moon.
to wait for the bugs to be worked out over a few releases. It's cool and awesome, but I don't know how useful it'd be in its present state of word-only translation.
cue the version that translates clothed women into naked women?
"LO TRADUCE EL TEXTO" is not a Spanish phrase, unless you want to say "Text translates it". The Spanish phrase would have been "TRADUCE TEXTO" but I think the result with that tool was so bad they changed the Spanish text until the bad translation rendered a good English phrase.
The same happens with other examples from that video such as "ROPAS OPCIONAL EN ESTA PLAYA". The only way you are going to read that sign is if you ask an English speaker to write it.
What they did was write the English phrases, translate them to Spanish and then translate them back for the video.
reminds me of a friend of mine who's hospital forced him to replace the typists that transcribed dictations with dragon speak software. They quickly discovered that it was pretty accurate except for small words like "not", as in "the tumor does (not) appear to be malignant"
translate geek-to-English? THAT would be magical!
$5 and it's not good enough!?!
On the other hand. On iTunes there's one review complaining that it should be free and ad supported. Sigh
If you don't risk failure you don't risk success.
The app is free but the Spanish-to-English functionality costs $5.
It is totally helpless vs. handwriting.
When viewing nice clean computer text on my screen, and when the phone is held very still, it produces the usual clunky translation.
It would be very much better than nothing when attempting rapid translations in a foreign culture.
FATMOUSE + YOU = FATMOUSE
No internet connection required for Word Lens. That's important for travellers for sure. Watch the video. They say that near the end.
What you say about cloud processing is true for things where an image is being compared to a massive data set (think Amazon's product identifier or Shazaam). In this case, the comparison is against a smaller data set, which is just two dictionaries worth of text and some linquistic rules. Not too challenging for a phone to manage. The burden that cloud processing avoids is the phone having to hold the huge data set and do large comparisons.
Where are all the people who feel it's not really necessary apply correct spelling ?
What a depressingly stupid machine.
No, this demonstrates what can be done when people write code using libraries that are compiled for a very specific hardware spec. Whereas Flash performs in a very un-optimized hardware-abstracted manner, it requires a lot of extra CPU to perform even the most rudimentary tasks.
You get back to me when this is implemented via Flash on Android and let's just see how long the battery lasts on your phone.
$5 / month hosted VPS on linux = awesome!
Is it good to have an instance app that gives me only one possible translation ?
Or is it better to get multiple once, like for example in this app here: http://itunes.apple.com/us/app/translatepro/id401440894?mt=8
Make it for Google TV and we can order our cartoons directly from japan, our dance films from India, and our porn from Germany...
A golden throne awaits whoever does this first.
The app does two things out of the gate that show it works - it has a mode that reverses words randomly, and another mode that erases random words. Both are damn impressive and show it can do what it says - so I went ahead and bought the spanish->English pack knowing I might have need of it in the future when travelling.
If you follow app sales closely (and being an iPhone developer, I take every chance I get to review trends) by far the way to make money in an app is with a free version that has in-app purchase. You have to change .99 (the minimum) and you literally get millions more people trying an application that is free than you do with an app that costs any amount of money. So I think they took the approach that will yield them far more money in the end, and also lets users have the language packs that are most useful to them.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Actually, Lo traduce el texto, meaning "it translates the text," is actually grammatical in Spanish; but only works in certain conversational contexts, and definitely not in one like that. It'd have to be something like this:
Despues que uno entra el texto, qué hace la aplicación? ("After you enter the text, what does the application do?")
Lo traduce el texto. ("It translates the text.")
This is of course of no credit to the writers of this "translation" app. It seems to be looping over the words, looking them up in the dictionary and spitting out the top translation it gets for them, with no attempt to actually take care of word order or to use the context in which a word appears to choose the best translation.
Are you adequate?
...not as nice as a globally decided and universally taught world auxiliary language would be (whether it might be English or whatever gained agreement). Would be much cooler not needing to use translation tools by actually having a shared language taught everywhere.
No Klingon?
Pleco Software ( http://www.pleco.com/ ) has a version of this for their excellent Chinese dictionary software. There's a video of the prototype at http://www.youtube.com/watch?v=x7VTo0656Rc
I'm not sure if the above works on the latest (4th-gen) iPod Touch with camera, or only iPhone.
I'm not affiliated with Pleco, other than as a very happy customer of theirs for about 8 years. I first got their electronic Chinese dictionary software for a Palm Pilot back then, and then more recently migrated (for free) to their iOS version for my iPod Touch. The dictionaries they license aren't cheap, but they're very good, and their software and support is great; I highly recommend them.
Most of the signs supposedly in Spanish make no sense and seem to be Google old automatic translations. Quite disappointing.
"Lo traduce el texto" is definitely not grammatical in standard Spanish, because it has two direct objects: "lo" and "el texto". It might be acceptable in some dialect I haven't come across.
Oh boy. This was one of the things I studied in my first year of grad school in Linguistics, so it was a long time ago. This is called "clitic doubling." I remember things being as follows:
Direct object clitic doubling in most Spanish dialects is permissible only in some conversational contexts, and IIRC depends on things like topic/focus structure of the dialogue, and parallel structure of coordinate clauses. The best examples I concocted in my research went something like this: A Pedro le mataron al hermano, a María le mataron a su madre, pero a Juan lo mataron a él. ("Peter, they killed his brother, Mary, they killed her mother, but John, they killed him"). In the final clause of that sentence, the direct object clitic doubling is in fact obligatory.
Are you adequate?
No, el texto is certainly the direct object. For example, the verb doesn't agree with it: La máquina traduce el texto. La máquina traduce los textos. Las máquinas traducen el texto.
Are you adequate?
Example: http://www.youtube.com/watch?v=A5Vb9SLkq5k from http://videosift.com/video/The-Manslator :)
Ant(Dude) @ Quality Foraged Links (AQFL.net) & The Ant Farm (antfarm.ma.cx / antfarm.home.dhs.org).
I've been researching this area for my masters. Just getting the basic text localisation (i.e. recognising an area as containing text) working reliably is very difficult - there are some good algorithms out there but in the real world, with 1000s of fonts, font sizes, angles, lighting conditions etc, I've yet to see a 1 size fits all approach. And even if you do find an area of text, throwing that into an OCR engine is going to produce garbage for the most part. In short, its quite easy to show something off in controlled conditions but I wouldn't expect anything like the performance seen in the video in the real world.
The above said, very impressed to see that on an iphone and for it to be so responsive; these things can only get better and once some form of viable HMD makes it onto the scene these types of application are going to be massive.
Some spanish boards were plagued with grammatical errors in order for the translation to work.
Examples:
"Lo traduce el texto" is wrong. It should be "Traduce el texto"
Reversal:
"Y lo va el otro direccion" is horrid. Should be "y va en la otra dirección" Although this is hard to translate without context.
Open Source Java Web Forum with LDAP authentication