Slashdot Mirror

← Back to Stories (view on slashdot.org)

Google Outlines AI-Based Number Reading For Street View Photos

Posted by timothy on from the may-not-combine-with-remote-detonation dept.

mikejuk writes "A recent Google research paper outlines how it might use AI to read digits in natural images — specifically Street View photos. The idea is to automatically extract the number of each house as captured by Street View and then use this to improve the geocoding data returned by Google. When you next ask for directions to a particular address the new data could be used to show you a street view looking directly at the house you specified."

68 comments

  1. what happens when it reads 0 / 0? by Joe_Dragon · · Score: 5, Funny

    does it crash?

    1. Re:what happens when it reads 0 / 0? by cultiv8 · · Score: 0

      It's in beta, they're still working out the bugs.

      --
      sysadmins and parents of newborns get the same amount of sleep.
    2. Re:what happens when it reads 0 / 0? by Ksevio · · Score: 4, Funny

      Might run into problems when it drives by my house at number:
      42'); DROP TABLE street_numbers; --

  2. US Census TIGER/Line shape files? by Anonymous Coward · · Score: 3, Interesting

    I've poked around the US Census data which has geographical coordinates of pretty much everything you can imagine - streets, natural landmarks, including addresses.

    How does most mapping software get its dataset? Why does Google need to find street numbers from the photos? (Probably because a public dataset like this isn't available globally) I imagine that Census data is a free alternative but professional geolocation data requires big fees.

    http://www.census.gov/geo/www/tiger/

    1. Re:US Census TIGER/Line shape files? by djl4570 · · Score: 2

      An underlying reason for this is to better correlate the map with the objects placed on the map. Accurately place the Pita Place on the map between the Jamba Juice and the Charbucks instead of putting it on the other side of the parking lot where there isn't a building.

    2. Re:US Census TIGER/Line shape files? by skids · · Score: 1

      At least some of the current store of street number data came from publishing HITS on Amazon Turk, so it's wetware-derived.

      --
      Someone had to do it.
    3. Re:US Census TIGER/Line shape files? by barry99705 · · Score: 1

      A lot of this comes from the R.L. Polk database. Those guys hire out at minimum wage. Would you trust that data? I know a few guys that sat at a Denny's for a week just reusing info from a phone book for their daily logs. This will by a lot more accurate, since I don't know too many people that put the wrong street number on their houses.

    4. Re:US Census TIGER/Line shape files? by chrismcb · · Score: 1

      Why does Google need to find street numbers from the photos?

      Google doesn't NEED to find the street numbers from photos. But evidence suggests that the data Google currently uses isn't quite accurate. Google thinks it can improve on that accuracy, for almost no additional cost.

  3. For once a development that isn't creepy by PSVMOrnot · · Score: 1

    Perhaps I'm not thinking hard enough, but this seems to me to be an development from one of the $InternetSocialMediaOverlords that doesn't seem creepy.

    I mean, this is a nice feature that will save time and be useful; but it doesn't go revealing yet more personal information, or infact anything that you couldn't do yourself by browsing streetview a bit.

    So yeah, a new shiny that doesn't yet make things worse or closer to 1984.

    1. Re:For once a development that isn't creepy by Surt · · Score: 1

      This improves the ability of the gestapo to avoid getting lost on their way to pick you up.

      --
      "Who is the Journal of Quantum Physics going to believe?" --Stephen Hawking
    2. Re:For once a development that isn't creepy by Anonymous Coward · · Score: 0

      In my opinion you hit the crux of the problem in your post.

      Yes it's possible to get this information by doing the leg work yourself, but that doesn't make it a good thing for people to have instant access to this information without that work. Law enforcement uses the same argument "we could get this information if we got off our asses and did the leg work" when trying to justify putting GPS units on suspect cars with no warrant, and we don't care for that very much...

      In general, people that don't have to work to gain knowledge or information don't appreciate the impact of it. In the US we make it far too easy, IMHO, to get a driver's license and a car and the result shows in how careless most of our drivers are. In recent history we have made large sums of money far too easy to get (in loans, inflated wages, stock games, etc..) and now we are a society that believes it is entitled to things that are beyond their means with little regard to how it will be paid for.

      Accepting any treads into our privacy for convenience is no different (IMHO) than accepting limits on freedom for the purposes of safety.

    3. Re:For once a development that isn't creepy by Time_Ngler · · Score: 1

      This isn't a development of the $InternetSocialMediaOverlords. It's just a bunch of nerds who know some algorithms and enjoy solving practical problems who thought this would be fun to attempt, and happen to work for the $InternetSocialMediaOverlords. Trying to look for a conspiracy in this would be akin to looking for a conspiracy in the way the leaves pile up in your driveway in the fall.

    4. Re:For once a development that isn't creepy by Anonymus · · Score: 1

      That's right, this isn't creepy, because StreetView itself is the creepy part. This is just refining it.

  4. AC's street number AI by Anonymous Coward · · Score: 0

    1. Did OCR report (with confidence >=95 percent) number N, where (0 <= N < 4000)? Choose N.

    2. Did OCR for property's left-hand neighbor report N, and for right-hand neighbor report (N+4)? Choose (N+2).

    3. Same as #2 with left-hand and right-hand reversed.

    4. Did OCR for property's left-hand neighbor report N, and for right-hand neighbor report (N+2)? Choose (N+1).

    5. Same as #4 with left-hand and right-hand reversed

    6. Did OCR for property's left-hand neighbor report N, where (N % 100) >= 80, and for right-hand neighbor report Q, where (Q / 100) <= 20? Choose (N / 100) * 100.

    7. Same as #6 with left-hand and right-hand reversed.

  5. Captcha just failed by Anonymous Coward · · Score: 2, Interesting

    If it can read a digit in a pattern then capchta just became useless as a way of keeping bots out of a website

    1. Re:Captcha just failed by djl4570 · · Score: 5, Funny

      Then we'll have to migrate to "Cutest kitten"

    2. Re:Captcha just failed by jdpars · · Score: 2

      Except that house numbers are generally made to be readable at a distance, whereas captchas are made to be readable up close, with scrutiny.

    3. Re:Captcha just failed by d4fseeker · · Score: 1

      Captchas are made to be unradable. The average human success rate at some Captchas is far lower than that of a OCR-based cracker.

    4. Re:Captcha just failed by Baloroth · · Score: 2

      You should look up how the captcha system works (or reCAPTCHA, anyways). It is digitally scanned old print (such as old editions of the New York Times). They feed it through several OCR systems, and the ones that it doesn't work on get thrown into the CAPTCHA system for humans to identify. This allows them to a) digitize massive amounts of old print material (using the humans interpretation of words that can't be read by computer) and b) ensure large sources of CAPTCHAs which are unreadable by present OCR systems. Material that can be read by the new OCR system simply won't end up as captchas.

      --
      "None can love freedom heartily, but good men; the rest love not freedom, but license." --John Milton
    5. Re:Captcha just failed by satuon · · Score: 1

      They can't use the words from old print because they don't know what it reads, so they can't know if you've answered correctly. The captcha that is used for verification is computer generated, that's why reCAPTCHA has 2 words - one is generated (hence they know what it is), and one is from old print. If you answer the computer generated captcha it's assumed that you've answered the other one as well. They also send the same old print word to 2-3 people to see if their answers match. That way Google gets a crowd-sourced human-brain-driven OCR machine for free.

    6. Re:Captcha just failed by justforgetme · · Score: 2

      you are unfamiliar with the tech underlying recatpcha

      why do you think it requests two words?
      one word it has classified one it hasn't.
      If you input the classified word correctly you pass. The input of the unclassified word is buffered.
      Unclassified words that get consistent results are added to the classified word pool and used as validators.

      It is a basic ml application whose computing is outsourced to the humans on the Internet who want to post
      on canHazMcRib.org and a very elegant solution as well. if you want more info on Luis von Ahn's ventures
      start with his ted talk.

      --
      -- no sig today
    7. Re:Captcha just failed by Ihmhi · · Score: 1

      Being pedantic here, but one of the words was probably read by OCR. One word in reCAPTCHA is unknown and the other is verified (but blurred, struck out, color inverted, etc.). It's likely that the machine-readable word was originally caught properly by OCR.

      --
      Random Thoughts From A Diseased Mind (Not For Dummies)
  6. captcha anyone? by Anonymous Coward · · Score: 0

    captcha anyone?

  7. Google is Skynet by itsybitsy · · Score: 2

    It's clear now.

    1. Re:Google is Skynet by Avarist · · Score: 1

      Skynet actually already exists. It's a Belgian company.

      --
      In Capitalist US, the commerce controls the Government.
  8. Street number reading by Animats · · Score: 2

    I've been wondering when they'd make that work.

    In retail areas, street numbers tend not to be too prominent. It may be necessary to read business signs and use that data to disambiguate addresses. This would help to clean up the phony-business problem in Google Places. An alternative is to use real estate records, as the USC Geocoder does for some areas, to get a solid lock on address vs. physical position. But that data is only available for some areas. There are also the Census Bureau's TIGER/LINE files, but they're US only and not complete for the entire US.

    Outside the US, this is likely to be more useful. If you have a few street numbers and a few business signs per block, you can infer the rest reasonably accurately.

    1. Re:Street number reading by ortholattice · · Score: 1

      In retail areas, street numbers tend not to be too prominent.

      I have found that to be true and don't understand it. While businesses loudly proclaim their names and logos, their street numbers are often barely visible if present at all. When dealing with heavy traffic, you often can't just leisurely slow down and rubberneck to search for some street number clue in an unfamiliar area. (Well, pre-GPS anyway.) More than once I've just given up trying to find a business while driving and gone to a competitor I was already familiar with. Is there some marketing reason that makes a street number detrimental to business?

    2. Re:Street number reading by Anonymous Coward · · Score: 0

      > Outside the US, this is likely to be more useful.

      Depends. In England (and I suspect this is true for the rest of the UK) you can locate about any address within 50 meters with the full postcode. However the house number is often absent (at least in London).

    3. Re:Street number reading by Animats · · Score: 1

      While businesses loudly proclaim their names and logos, their street numbers are often barely visible if present at all.

      That's true, and annoying. Some cities require street address signs but most don't.

  9. I have an idea.... by salva84 · · Score: 4, Funny

    http://cache.gawkerassets.com/assets/images/4/2011/11/8ca4aa827c1151851324895f5ffb778b.jpg

    1. Re:I have an idea.... by mapkinase · · Score: 1

      I wonder how passing like this through unsanitized forms works. The final SQL statement will include unclosed quotes and would produce an error.

      --
      I do not believe in karma. "Funny"=-6. Do good and forbid evil. Yours, Oft-Offtopic Flamebaiting Troll.
    2. Re:I have an idea.... by mapkinase · · Score: 1

      Nevermind, saw the comment in the end.

      --
      I do not believe in karma. "Funny"=-6. Do good and forbid evil. Yours, Oft-Offtopic Flamebaiting Troll.
    3. Re:I have an idea.... by jheath314 · · Score: 1

      Did you really get the city to change your house number to 404'); DROP TABLE Streetnames;-- ?

      --
      Procrastination Man strikes again!
  10. Other uses by Anonymous Coward · · Score: 0

    If its able to read the digits in all those varieties, then it will probably be great at reading captchas.

  11. Is this the paper Norvig mentioned? by Obispus · · Score: 1
    Is this the paper Peter Norvig (quite tangentially) referred to near the end of this article in the MIT Tech Review? Can anybody confirm?

    Thanks...

  12. AI required. by Anonymous Coward · · Score: 0

    For an advanced Tolitarian state to perpetuate a surveillance society, an AI must exist which is powerful enough to search the infinite amount of data created, and determine it's enemies as efficiently as the market determines winners and losers.

    1. Re:AI required. by RockDoctor · · Score: 1

      For an advanced Tolitarian state

      Is that an oxmoron? Possibly. But I'll let it pass for the moment.

      to perpetuate a surveillance society, an AI must exist

      MUST ? I really hate people who use excessive didacticism ; I think that they should be taken out and shot in front of Jeremy Clarkson's family.

      which is powerful enough to search the infinite amount of data created,

      Consult your dictionary for the difference between "large" and infinite. If you feel brave, or are already under medical care, you may wish to read Georg Cantor's versions of the Necronomicon.

      and determine it's enemies as efficiently as the market determines winners and losers.

      OIC, it's all a joke. Sick fuck.

      --
      Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
  13. I wonder by koan · · Score: 1

    If the proliferation of CCTV using facial recognition could be tied into Google allowing you to Google for faces and where they were last seen.

    So using a 3 Dimensional recognition algorithm and saving that data plus the location would then allow you to drag a picture of your friends face into Google (as you already can) then get results of where that face was last seen.

    --
    "If any question why we died, Tell them because our fathers lied."
  14. Isn't this a good job for the mechanical Turk? by wisebabo · · Score: 1

    Could humans, with their supremely evolved optic nerve/brain, outperform any A.I. that Google might have now or in the near future? Might it even be cheaper? (Just how much does difficult A.I. cost in the cloud anyway?). Maybe even if A.I. is much cheaper, it would still be useful for using humans for the difficult cases, or as error correction. Best would be for the humans to train the A.I. (and themselves out of a job!)

    I always thought this was the only possible reason how the machine intelligences in "The Matrix" could use humans. While they lay in their pods, dreaming their lives away, use the unused portions of thir brains (no, not 90%) to perform tasks that humans are "better" at. Like image recognition or perhaps language. That whole "battery" idea was just stupid.

    Then again they should've had their eyeballs plucked out and a connector wired directly to their optic nerve. Kinda would've made making a movie, in which Neo and all the other humans were stumbling around blind, difficult.

    1. Re:Isn't this a good job for the mechanical Turk? by RockDoctor · · Score: 1

      I'm sure that it could be done by the Mechanical Turk. But considering the amount of data that they've got ... possibly it would be too expensive. The Turk may be flexible and relatively cheap, but it's still got a finite cost per computation, and meat-puppets are not particularly cheap. (It might be useful for cross-checking hit rates etc though. Quicker and cheaper than coding and testing a second/ third/ fourth algorithm in detail.)

      --
      Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
  15. Last lesson of ml_class.org by doudou42 · · Score: 4, Interesting

    It's interesting to note that one of the co author is also the teacher for stanford free classe on Machine Learning and that the last lesson of the course was on this topic...

    1. Re:Last lesson of ml_class.org by Mr.+Shotgun · · Score: 1

      You noticed that to? It's nice of him to pass along some of the stuff he is working on as a lesson for the class.

      --
      Of all tyrannies, a tyranny sincerely exercised for the (supposed) good of its victims may be the most oppressive
    2. Re:Last lesson of ml_class.org by w_dragon · · Score: 1

      Don't most profs do that? Isn't that the whole point of having PhDs doing both research and teaching? I know most of my 4th year profs talked about what they were working on in class...

    3. Re:Last lesson of ml_class.org by RockDoctor · · Score: 1

      nice of him to pass along some of the stuff he is working on as a lesson for the class.

      Almost as immoral as using the Mechanical Turk! (As someone else suggested upthread.)

      --
      Birds are not dinosaur descendants;birds are dinosaurs, for all useful meanings of "birds", "are" and "dinosaurs"
  16. oh by negrace · · Score: 0

    what if I live at Sales, 20?

  17. I did this manually for X-mas cards by Anonymous Coward · · Score: 0

    I feel like a high beyonder stuck in the slow zone, damn you Pham and your disaster!

  18. Captcha by numbscholar · · Score: 1

    Will it work with those annoying recaptcha boxes?

    --
    GENERATION 26: The first time you see this, copy it into your sig on any forum and add 1 to the generation.
  19. It's not AI. by forkfail · · Score: 1

    It's machine learning.

    There's a difference.

    --
    Check your premises.
    1. Re:It's not AI. by sanchom · · Score: 1

      It's machine learning.

      There's a difference.

      Please explain the difference, then. At Stanford, UBC, MIT (I assume other places as well), their artificial intelligence labs include their machine learning groups.

      For example, at Stanford, the Artificial Intelligence Laboratory includes the research groups of Daphne Koller, Andrew Ng, and Sebastien Thrun, who all describe their research as Machine Learning.

      Andrew Ng, a machine learning research and instructor for the online machine learning course (http://www.ml-class.org/course/auth/welcome) is director of the artificial intelligence laboratory at Stanford.

      If you say the techniques in the article are machine learning, then they are also artificial intelligence. The original post title is correct.

    2. Re:It's not AI. by Anonymous Coward · · Score: 0

      Or you could Google it

      "Machine learning, a branch of artificial intelligence, is a scientific discipline concerned with the design and development of algorithms that allow computers to evolve behaviors based on empirical data, such as from sensor data or databases. "
      http://en.wikipedia.org/wiki/Machine_learning

      I bet you would only have to tell an AI once to Google it.

    3. Re:It's not AI. by sanchom · · Score: 1

      I did google it. That's how I found out the info in my post, and I also the wikipedia article you linked to. The wikipedia article says, as I did in my post, that machine learning is a branch of artificial intelligence. So if you say something is machine learning, it is also artificial intelligence.

      Calling this work "AI-based" is correct.

      The amount of times you'd have to tell an AI something in order for it to learn is irrelevant to this discussion.

    4. Re:It's not AI. by Anonymous Coward · · Score: 0

      I disagree for 2 reasons, AI or Artificial Intelligence is the attempted simulation of human intelligence and therefore does not exist at all currently, machine learning by its self can not be considered "intelligent" it is a facet of intelligence.
      If anything I would say that the term AI is overused and overused incorrectly, there is nothing intelligent about machines (currently) the term most often gets used when comparing a "dumb" machine, one that can not adapt if it's situation or parameters change to "intelligent" machine that can, both are still programming and involve little to none of the characteristics of true intelligence seen in humans, in fact there is no single definition of human intelligence as no one can quite agree on an exact grouping of facets that define it.

      We all know it when we see it though.

  20. It's not as scary as it sounds by Jason+Z.+Christie · · Score: 0

    After all, your address is public, until it's tied to your name. Even that info is more or less public in a lot of cases (name on the mailbox?) But it does open up the potential for fun new hacks and pranks... It'll be a good test of AI and OCR, ultimately. I sense new captchas on the horizon.

    --
    Zombie Killa, Free nerdcore hip-hop novella http://www.smashwords.com/books/view/119619
  21. there's a famous street sign in china by circletimessquare · · Score: 1

    they were trying to appeal to english speaking olympics visitors in 2008, and the translation server crashed, but they thought the error message was the translation, so we get:

    http://boingboing.net/2008/07/15/chinese-restaurant-c.html

    so would it be like divide by zero if machines try to map the real world and encounter a bit of the real world mimicking machine world fail?

    --
    intellectual property law is philosophically incoherent. it is your moral duty to ignore it or sabotage it
  22. Eh? by MrL0G1C · · Score: 1

    What the f*** would the point of that be? in the UK they blur out door numbers and street signs which is bloody inconvenient when you're researching visiting a place.

    --
    Waterfox - a Firefox fork with legacy extension support, security updates and better privacy by default.
  23. Hacking the system: by drolli · · Score: 1

    You open a shop in the same road as a very sucessful one and put their number sign (and the shop name, in case they use OCR on it) just when the google car passes by......

  24. Sponsored by the US drone program by Mister+Liberty · · Score: 3, Funny

    n/t

  25. Augmented reality meets google maps by Anonymous Coward · · Score: 0

    This might path the way for some nice augmented reality gear.

    Look at a building though your augmented gear/apps and see that there is a Japanese restaurant inside the building and it is rated 3 stars.

  26. Now the big question remains..... by Anonymous Coward · · Score: 0

    can you opt out?
    Or are they now going to show everything they can collect on an adres when you punch it in?
    Here is also the facebook, linkedin, email adres, etc.... of the persons of interest on this adress

  27. The number most frequently seen by Harald+Paulsen · · Score: 1

    is 867-5309

    --
    Harald
  28. I'm sure the military has already worked this out by Trax3001BBS · · Score: 1

    I remember when I first heard that the spy satellites could read a cars license plates.

    I thought that's good trick, not only do they have to view small characters they
    had to do it sideways.

  29. Licensing issues, more likely by msobkow · · Score: 1

    In Saskatchewan, we have a Crown Corporation that is responsible for maintaining, updating, and distributing that map data for the province. It's accurate data -- collected using actual transits and GPS systems. Google can not improve on the data quality by reading street signs with AI at all.

    I suspect it has a lot more to do with licensing issues for Google wanting to republish the data to the general public, something that I'm sure our crown corp would frown upon as it would cut out their geodata sales market. Google wants a royalty-free data source, not better data quality.

    --
    I do not fail; I succeed at finding out what does not work.
  30. Or they could team up with OpenStreetMap by olau · · Score: 1

    ... and get an army of volunteers to help them.

  31. StreetView is very low-res by LanMan04 · · Score: 1

    Most StreetView images are soooooo low res that you can't make out street numbers on most houses.

    Maybe businesses with large street number signs, but not houses.

    --
    With the first link, the chain is forged.
    1. Re:StreetView is very low-res by Anonymous Coward · · Score: 0

      That is certainly true from the front-end, but I would be willing to bet that Google has high resolution images and simply scales back the quality to improve the performance of the web interface. I have a hard time believing that they would go to the trouble of driving down an appreciable portion of the world's streets without putting a decent amount of money into the camera gear. That also allows them to upgrade street view as average bandwidth and device speed increases allow without having to drive all over the place again. So in response to your post, I'd say that no, you can't make out street numbers but they most probably can.