Catch up on stories from the past week (and beyond) at the Slashdot story archive


Forgot your password?

Point, Shoot and Translate into English 159

edstromp points out this New York Times "story on using a pocket pc to translate a street sign. It requires at least a dialup connection as it sends the photo to a server for the majority of the processing: OCR, translation, English overlay for new image, and then transmission back to the user. All said and done, it takes about 15 seconds to translate a street sign. Put this with some augumented reality, and you have a rather useful tool."
This discussion has been archived. No new comments can be posted.

Point, Shoot and Translate into English

Comments Filter:
  • by crumbz ( 41803 )
    I can't wait until that is embedded in a contact lens.
  • ...while driving.
  • by mikosullivan ( 320993 ) <miko.idocs@com> on Thursday March 14, 2002 @11:03PM (#3166367)
    You walk down the street and the cute girls are modified so that they appear interested in you.

    ... oh wait, we already have beer for that.


    • I was thinking more like you walk down the street and all girls are modified so they appear cute... ...but beer does that too...
    • I though beer just made the girls who are interested in you appear cute?
    • Check this out .. Lovege [].
      TOKYO -- On the streets of Shibuya, Shinjuku and other popular teenage hangout districts in Japan, the days of composing a provocative pick-up line and mustering the courage to utter it smoothly are on their way out. Young lovers are letting Lovegety do the talking for them. It was only a matter of time before the male and female Tamagotchi, toys that hook together to breed kids, would spawn another animal. And this one is at least as popular as its ancestors. "In the two and a half months since the product entered markets, we've already shipped 350,000 Lovegetys. Right now we can't produce enough of them to meet demand," says Takeya Takafuji, executive of Erfolg, the company that manufactures Tamagotchi and Lovegety. Affectionately called the "Lovege," the oval device has three buttons the user sets according to the kind of activity she or he has in mind: "talk," "karaoke," and "get2." (The latter is a wildcard -- but perhaps could mean "get to it.") Once the holder selects a mode, the device searches for Lovegety holders of the opposite sex in a five meter radius. If it locates a holder with the same mode, the "get" light flashes and the device beeps, so the pair can find each other.
  • This can't be too far away, then. in Michael Crichton's "Timeline," they have little earbuds that can translate spoken words into something that the wearer can understand, on the fly. If only we had something like this in reality. Not too much farther away...
  • I believe Linux Conf Australia had a speaker (see h) who discussed converting predefined glyphs into 3D overlays. Extending this capability to a wider set of real-world signage would be a worthy project for the Linux-based PSX2 and it would require insane amounts of image processing.

  • Go to and click the link on the right side of the page that says "Does your PDA parler français?". It is video for a translation device. It's pretty amazing.

    The guy was talking into it in English and this thing repeats the words in the selected language.

    I'm sure it's far from perfect, but this thing is like one step closer to some Star Trek like technology in regards to translation.
    • Does your PDA to speak french? They could have at least used the right conjugation for 3rd person present tense: parle.

      I refuse to accept such a level of American-centric reporting, especially when it contains such blatant grammatical errors as these. Bah!

      • What's the problem? That's perfectly valid franglais.
  • i can see it being useful if you have some way of connecting, but without it, all you will get are some pictures of signs in a foreign language.
  • by Space Coyote ( 413320 ) on Thursday March 14, 2002 @11:05PM (#3166380) Homepage
    ...Would be a similar system that can use OCR to read street signs and then send the text to a voice synthesyzer. Seems like that would be endlessly useful for people with low vision who have trouble reading signs in awkward locations.
  • by TrebleJunkie ( 208060 ) <ezahurak&atlanticbb,net> on Thursday March 14, 2002 @11:06PM (#3166385) Homepage Journal
    Combine this technology with last summer's craze,, and I think you got something. (15 seconds to know if the chick you wanna pick up in a bar is really hot? Priceless.)
  • by Anonymous Coward
    Went to one them countries abroad 'while back. Filled with foreigners, it was. Everywhere I looked was foreigners yammering in some strange foreign tongue. I can't see why they just don't learn to speak English...much easier than wasting my time with some sort of pocket translater.
  • by Anonymous Coward
    Think of all the better things they could do. They could translate it from Enligsh to hax0r! "5p33d l1m17: 55 mph" "570p!" "j00 c4n 4d0p7 4 h1ghw4y!"
  • How long... (Score:2, Funny)

    by theMAGE ( 51991 )
    How long till' people will drive using this as input? "Computer: what is that red sign over there?" ... crunch scan crunch ocr crunch exception: macromedia plugin required crunch downloading... ... 15 seconds later, from the car's wreckage: IT'S A STOP SIGN. REPEAT IT'S A STOP SIGN
  • by willybur ( 217434 ) on Thursday March 14, 2002 @11:07PM (#3166393) Homepage
    On that note, this month's issue of Scientific American features an article on augmented reality []. It's a good read.
  • Driving down the street...

    What's that sign?

    *click* (take a photo)


    Translation comes in: "STOP" sign
  • ...just imagine visiting some far-away place, sending of a picture of a street-sign for translating, and getting back "beware the polar bears"!
  • Suspicious (Score:2, Funny)

    by EvilBastard ( 77954 )
    I was messing with the Prototype for this, and after I tried about the 40th sign, I got this back

    Warning !
    Wrong Way
    Go Back

  • Definately reminds me of the babelfish... although I don't think they worked for reading... I read an article a month or so ago about a research group working on AR (augmented reality). The setup they had weighed about 50 pounds and required a big bulky headset. But with the rate of miniturization these days... who knows what'll be possible in the not-too-distant future. I envision everyone wearing glasses that project the AR, and connect to a computer in a pocket somewhere... Maybe it'll be contacts.
  • Lazy lazy! (Score:1, Interesting)

    by Aurorya ( 557733 )
    Part of the joy of going somewhere (for Americans) other than America is that everything is probably in a foreign language. And that people refer to these places, concepts, foods, etc., in the language that describes them best: their native one. I'm not going to go to la Rue d'Auguste Lançon in Paris and have my cute little pocket device translate it into "August Lancon Street". Parisiens giving me directions are going to call it la Rue d'Auguste Lançon!

    Ok, so what about China or Japan? If you are going for travel, you can learn a few Kanji. It's the least you could do. If you're going on business, as the article suggests, you should be a good little representative and be chosen because you know something about where you're going. Hopefully you know a lot, or at least enough to be able to order food from a menu.

    It's kind of sad that no people won't even have to make the smallest step into being somewhere new of calling places by their real names. If you lovingly name your kid George, would you be upset if the Mexicans only refered to him with the pronunciation "hor-hey"?

    • Re:Lazy lazy! (Score:1, Insightful)

      by Anonymous Coward
      I agree with you when you're in Europe, but it's a different story in places like Taiwan, where I'm working at the moment. There are some signs in Latin text, but some of the minor roads are only written in Chinese.
      When someone tells you to go to Chong-Shen North Road, how would you know that äå±±åOE--è is the correct address? I would love a gadget like this! It would make living here so much easier.
      It would make learning Chinese easier as well.

      (Apologies if the Chinese doesn't display properly).

      Col the Limey
  • Cached (Score:3, Informative)

    by BrianGa ( 536442 ) on Thursday March 14, 2002 @11:14PM (#3166426)
    Google [] has a cache [] of the PopSci article [].
  • by KingPrad ( 518495 )
    ...can you get through the airport with it? Carry any more technology and those security guards will tear you apart.
  • For all those who need one:

    l: slashdotter1
    p: slashdotter1

  • Well, this is about as useful as my flying toaster!

    Seriously though, i wonder about what happens when the sign gets babelfished and you end up eating dog testicles instead of the "beefy camel tips" you thought you ordered.
  • by cporter ( 61382 ) on Thursday March 14, 2002 @11:22PM (#3166447)
    I don't see this as a positive development. It's an excuse for crappier signs in a world where signs, schedules, maps, and notices are confusing even if you're fluent in the language. We should focus on standards [] and intuitive design. []

    OK, maybe translating train schedules and restaurant menus is good. But street signs, especially, are supposed to be unambiguous, their meaning readily apparent to anyone, whether literate in their native language or not.

    And does this thing work on signs that some redneck has shot holes in with a 12-gauge?

    • What are you talking about? You want unambiguous, you use longitude and latitude. "Yeah, I'm only a few seconds away from 40 degrees latitude and a minute from 43 degrees longitude" :)

      Flame me for not knowing where 40x43 is :P Its a joke son, ya' hear'?!
      • The latitude and longitude you gave could correspond to 4 different locations on earth, so I'd say that's pretty amibiguous too. =]

        Although, if you go with 40 N 43 E, it's approximately somewhere in the middle east. (Some people use a +- system, searching this was the first time I had seen that, I've only used N S E W, etc.)
      • He means things like stop signs, yeild signs, pedestrian crossings, crap like that, not signs that give the names of streets.
  • Configure the translator as a web service with a fairly-obvious SOAP API. Do everything in Unicode. Allow a wide range of both input and output languages. Make plenty of clones of the server, sprinkled all over the planet. Start in tourist spots and big cities, and let them trickle out to other places.

    In the 1950s and 60s, TV commoditized and homogenized American speech patterns and culture. This will commoditize understanding between cultures, but nobody has to give up their native language. Ideas and commerce will flow more easily. It'll be a good thing.

    • In the 1950s and 60s, TV commoditized and homogenized American speech patterns and culture.

      Do you thing this is A Good Thing (tm)? (This is not ment as a flame, but food for thought, but on second thought, maybe it is a flame.)

      Disclaimer: I'm not from the US of A, but from Europe and I really like finding a different culture every 100 kilometers (or miles).

      • Do you thing [the homogenization of American culture in the 50s and 60s] is A Good Thing?

        I didn't say that. We don't need to homogenize world culture for this thing to work. What gets homogenized is understanding of other cultures. Each person stays within the dialects and habits of his or her own culture, but sometimes learns a little bit about others.

        Maybe if there had been something like this for America in the 50s and 60s, a Texas/Maine translator say, we wouldn't have ended up with our cultural homogeneity today. Though really, neither folks from Maine nor Texas have made many concessions to cultural homogenization.

        The real evil of American cultural homogenization, such as it is, is the influence of big corporations. They'd benefit by commoditized cultural understanding, but individuals would benefit even more. So I don't see this as a call to arms for the Dark Side.

  • Somewhat Related... (Score:3, Interesting)

    by peterdaly ( 123554 ) <petedaly&ix,netcom,com> on Thursday March 14, 2002 @11:34PM (#3166482)
    On a somewhat related topic I have been thinking about recently:

    Java (J2ME) is now in cell phones, I have one and have played around a bit. Biggest problem with real applications is lack of a good input device. Now, for speed dialing, my phone has "voice recognition", which is really a pattern match against a saved database of me saying each person's name. It is an i85 Nextel Phone.

    Why not have a voice recognition processor? Now, the phone does not have enough horsies to crunch the stuff needed to do that...but: The phone has direct-connect. Why not a feature like direct connect, but instead of 2-way radioing another person, a voice processor system, which returns the processed speech as text into whatever is running on the phone? Take the time used out of alloted's not like they have to connect anything in your call to the phone system to establish a call for you.

    Data connection is only about 300 baud or so, but how much faster can you really talk (so that a computer can uderstand you) than 300 baud worth of text? Same thing for reading. I can't read my email while driving (at least not safely), but why not have a "my phone" (really a computer talking to my phone) read it to me? That solves the small screen display problem too.

    Ok, enough crazy thinking for now, I could go on and on about this stuff.

    • You can get 100% java voice recognition systems, so you could maybe do it on one of the new mobiles.

      Still does not help for Quake, but for MUD's it would be cool...

      But all that said and done, it would be easy to a system that you phone up your PC, it answers and send what you say through a voice rec. app, the pipes the output to a shell.

    • Data connection is only about 300 baud or so, but how much faster can you really talk (so that a computer can uderstand you) than 300 baud worth of text?

      If your phone were doing the translation, that might be true. But combine that statement with one you made earlier:

      Why not a feature like direct connect, but instead of 2-way radioing another person, a voice processor system, which returns the processed speech as text into whatever is running on the phone?

      ... and you've got a lot more than 300 baud of information being sent. If your phone knew enough to filter it down to 300 baud, it wouldn't need to send it at all.

    • Well some operators already have this service.

      In sweden both Europolian (soon to be Vodaphone) and Telia have voice services for internet.

      Its all configurable through the web, If you want a machine rings you and reads up each mail as it is recieved
    • theres new low power chipsets that allow you to do voice recognition processing in small devices. it all comes down to cost, power consumption (they're designed for low power devices aka handhelds,phones etc) and trainability of the chip software - aka usability.

      in other words, its probably more likely that you'll see the functionality rolled into your pda or phone.
  • So if somebody flips you the bird, take a picture and find out what he was really saying!
  • by suso ( 153703 ) on Thursday March 14, 2002 @11:43PM (#3166517) Homepage Journal
    What do you do if you're trying to read signs that would lead you to a cybercafe?
  • But why not just have a bluetooth type chip embeded into all signs and landmarks? Walk up to it with some type of reciever/pda/device and wha-la. Even embed a gps etc.
  • This all relies too much on sight. These things, no matter how efficiently designed, do get in the way of your sight, even if transparent. Why use sight at all? Personally, I would prefer a system with a larger learning curve but effectively bringing a new sense into play. Electrodes applied to the body would send raw singles, which we would learn to interpret. This could have many possibilies. "Oh, would you look at that dream-woman, my alarm is going off. Hold on, let me set it back an hour." All while sleeping. Well, maybe not, but I think its cool. Just an idea.
  • I'm suprised (Score:2, Informative)

    by madenosine ( 199677 )
    I'm suprised that nearly nobody has carried the story of the brown reaserchers who put a microchip into a monkey's brain which allowed them to control a computer mouse by thinking

    They first played a game with a joystick, then played the same game controlling it with their mind, and they got about the same score both ways

    Very interesting story.....has anybody seen anything on this? It's on brown's website at 01-02/01-098.html []
    • Now this is what im talking about. I want this! Do you think there are any legal issues against doing this to myself? And does this work both ways? ie, could the joystick in their head have "force feedback", therefor applying a new sense?
    • I posted that story here around noon today, but it was rejected. I find this sort of thing very intriguing, interfacing machines to neurons and such. There's crazy potential to help people, and it's freaky futuristic.

      What I want to see is a video of these monkeys, to see how accurate it really is (which I'm sure also depends on the monkey's intentions as well).
  • I'm on a business trip in Paris. My wife flys in to be with me. Fourteen hours later I show up at the airport.

    Wife: "Where the hell have you been!!!"

    Me: "server was slow..."
  • Put this with some augumented reality, and you have a rather useful tool.

    ... Just stay out of Canadian Airports, eh?

  • Does it do American to standard English road sign translation? You know... something for the tourists:

    <STOP>: "Prepare for car-jacking"
    <DRIVE-THRU>: "Drive By"
    <WELCOME TO LOS ANGELES>: "Welcome to HELL!"

  • New York Times new slogan: "All the news fit for nerds"

    Slasdot's: "News from New York. Stale links that mattered 6 hours ago."
  • These augmented street signs would be incredibly cool if they showed up as small popup boxes on your field of vision as you're looking for streets. Better yet, augment with mapquest, and have a line to follow to get to your destination. Adjusting for traffic and the ilk. That would be cool.
  • i've trained myself to do similar things without computers. i was enrolled in a state funded training program for 12 years, where they taught me to identify english road signs, read them, translate them into english, and understand them. i also learned how to translate the following things into english:
    • books written in english
    • product labels written in english
    • movie subtitles written in english
    • magazine articles written in english
    then, after my 12 years of state funded black-ops training, i decided to continue my education in a private 4 year institution where they even taught me how to WRITE in english, a topic which wasn't quite covered well by the state-funded institution. perhaps if the government would make this type of training available to all members of our society, we wouldn't need computers to understand these cryptic road signs that nobody seems to be able to decode.

  • While driving along some roads...

    Take a picture of that sign and see what it says.

    -15 seconds later-

  • Improvement idea (Score:2, Insightful)

    With the processor speeds and storage abilities of current Pocket PCs, why require a dialup connection? Especially when the majority of users wouldn't have one while, say, traveling in a foreign country. (GSM still has yet to pick up here in the US).

    Why not allow the OCR program, and any necessary foreign language translation dictionaries to live in the PDA's memory? I can't see it taking up anymore than several MB, (which could certaintly be offloaded when not traveling).

    Is there something I really don't understand here?

    • Yeah, that's all fine and dandy, in fact some might even say you're making sense. But gotta remember, they are going to want to generate revenue. This means ads and/or subscription fees. That's why they want you to go through them.
  • It requires at least a dialup connection ...

    Not bloody useful when I'm walking down Broadway, is it then? (rimshot)

  • Last summer I created a prescription delivery system that received a faxed prescription (the Rx was faxed from another company's software to a Pharmacy; in our case the "pharmacy" was a fax-to-email company in Vancouver, B.C.) as a TIFF attachment to an email. Then I parsed the email, scanned the distinct areas (well-defined, thankfully) of the fax for the details of physician, patient, prescription, sliced the fax into pieces and reformed those pieces to create a label to be printed on the drug to be dispensed. Then I popped the Rx to the correct awaiting browser (real-time dispensing) and an authorized user OK'ed the print and dispensed the drug.

    It took 2 weeks and used all open source tools:

    • Perl
    • Apache
    • Linux
    • Fetchmail
    • Mime::Parser
    • JOCR (was Gnu OCR but renamed for SourceForge)
    • other stuff
    Of course, that was for proof-of-concept: I then redid the application using SOAP::Lite for receiving an XML payload with the same data.

    It was a load of fun and proved to me that CPAN, SourceForge, Freshmeat, and Google are the only tools I need to get stuff done on a grand scale

    Regarding JOCR - it's not OmniPage by a long shot, but for specific OCR needs is worth looking at.

  • Cliff's been doing it for years. [] This is good stuff, my friends... and you get to fulfill your Linux and individuality dreams. What could be cooler?
  • Try to wear it through airport security! (The last thing your LCD goggles/speakers report is "Pleeze come this w4y to thiz small priv4t3 roam so that we might radish you." Uhoh.)
  • I know this is a little bit off topic, but while the technology is cool, you'd get some really kooky translations if you went around some foreign country translating the signs. I used to live in Japan, and after I learned a little of the language, I started thinking that the Japanese have very odd naming schemes. My house was loaced between the train stops "Cherry Blossom Palace" and "Happy Island" and I lived on the street "Middle of the rice field" right across from the megamart "Big Circle". Needless to say, there were neither palaces nor islands anywhere in sight, and the closest thing to a "field" was the parking lot, and I never saw any rice there... I did see some circles, but none of them were particularly large :) Still, I'll bet it would be great for reading menues in korean resturants and finding out just which part of the cow that last slice of beef came from... or I HOPE it was a cow!
  • by Sivar ( 316343 ) <charlesnburns[NO@SPAM]> on Friday March 15, 2002 @03:06AM (#3166993)
    "Dear, what does that sign over there say?"

    [15 seconds later]

    "It says: 'Road ends: Bridge constru...'"
  • "Hey, what does that one say?"

    "Hold on, I'll check."

    (a few seconds pass)


    "Um, this says it was a stop sign."

  • ...with a picture of a beautiful, but clothed, member-of-the-appropriate-sex.
  • Just tack on a voice synthesizer and it could become a useful tool for the blind. They would be able to hear the words written around them.
  • by anonymous loser ( 58627 ) on Friday March 15, 2002 @09:28AM (#3167559)
    "The translation service is a great application of augmented reality," said Dr. Seth Teller, an associate professor of computer science and electrical engineering at the Massachusetts Institute of Technology. "I had a terrible time finding my way through the Tokyo train station last month."

    Just about every meaningful sign in Tokyo station is already translated into English. In fact, there are few train stations within a 50-mile radius of Tokyo that don't have English language signs, at least for the essential stuff (this way to Harajuku, etc.). It's only when you start getting out in the country that reading signs becomes a problem for English-speaking foreigners. E.g. most stations on the Meitetsu line outside of Nagoya completely lack English-language signs. As a tip for foreign travellers in that situation I offer the following advice: follow the crowd. You are pretty much guaranteed by natural law to end up in the city center.

  • i got a device that fills your requirements. it tells me what street i'm on, what street i'm approaching and even what time it is! and it doesnt even need to be able to see the street signs. of course, it needs to be able to see at least 3 of its friends to know exactly where i am, but that isnt too hard when you're outside.

    i of course am talking about my GPS.
  • It seems like a more useful tool, would be a standard translator, where you could type in the words, and get the equivalent in your target language. Verb conjugation would also be very important. Franklin / Larousse makes some decent translators, but they are a little too large, and don't hold up very well. Does anyone know if there is a port or similar product for handheld devices?

  • We already have augmented reality devices, it is called "beer". It makes every woman look good and every guy your best friend/enemy.

These screamingly hilarious gogs ensure owners of X Ray Gogs to be the life of any party. -- X-Ray Gogs Instructions