Want to read Slashdot from your mobile device? Point it at m.slashdot.org and keep reading!

 


Forgot your password?
Close
typodupeerror
×
AI

Famous XKCD Comic Comes Full Circle With AI Bird-Identifying Binoculars (arstechnica.com) 70

Posted by BeauHD from the nine-years-later dept.
An anonymous reader quotes a report from Ars Technica: Last week, Austria-based Swarovski Optik introduced the AX Visio 10x32 binoculars, which the company says can identify over 9,000 species of birds and mammals using image recognition technology. The company is calling the product the world's first "smart binoculars," and they come with a hefty price tag -- $4,799. "The AX Visio are the world's first AI-supported binoculars," the company says in the product's press release. "At the touch of a button, they assist with the identification of birds and other creatures, allow discoveries to be shared, and offer a wide range of practical extra functions."

The binoculars, aimed mostly at bird watchers, gain their ability to identify birds from the Merlin Bird ID project, created by Cornell Lab of Ornithology. As confirmed by a hands-on demo conducted by The Verge, the user looks at an animal through the binoculars and presses a button. A red progress circle fills in while the binoculars process the image, then the identified animal name pops up on the built-in binocular HUD screen within about five seconds. In 2014, a famous xkcd comic strip titled Tasks depicted someone asking a developer to create an app that, when a user takes a photo, will check whether the user is in a national park (deemed easy due to GPS) and check whether the photo is of a bird (to which the developer says, "I'll need a research team and five years"). The caption below reads, "In CS, it can be hard to explain the difference between the easy and the virtually impossible."

It's been just over nine years since the comic was published, and while identifying the presence of a bird in a photo was solved some time ago, these binoculars arguably go further by identifying the species of the bird in the photo (it also keeps track of location due to GPS). While apps to identify bird species already exist, this feature is now packed into a handheld pair of binoculars.
This discussion has been archived. No new comments can be posted.

Famous XKCD Comic Comes Full Circle With AI Bird-Identifying Binoculars More

Famous XKCD Comic Comes Full Circle With AI Bird-Identifying Binoculars

Comments Filter:

  • First there was an XKCD for that... (Score:1)

    by Anonymous Coward

    ...and then it happened. Life is good.

    • Re: (Score:2)

      by znrt ( 2424692 )

      not really, bc the original vignette doesn't remotely suggest binoculars that identify bird species. this is just more covert advertisement with lame clickbait for nerds because (breaking news ahead) nerdity and stupidity aren't mutually exclusive. is that good?

      • because (breaking news ahead) nerdity and stupidity aren't mutually exclusive. is that good?

        Well, sort of, yes.

        Another reminder )for anyone with the mental flexibility to handle it) that people in pretty much any group (not categorised on IQ-ish measurements) can be stupid. And by corollary, different people in those same groups can also be intelligent.

        Yes, that does mean that there could be intelligent Nazis. Or even, to prod some current sensibilities, Nazi-like Israelis.

        Perceived intelligence is not a

  • Full Circle? (Score:4, Informative)

    by YetAnotherDrew ( 664604 ) on Wednesday January 17, 2024 @05:41AM (#64166341)

    I don't think Benj Edwards at Ars knows what "comes full circle" means. Alternately, someone else chooses Benj's headlines and has trouble with English.

    And no, I don't mean whatever activities I'm sure responses will mean when they talk about "Benj coming full circle." I can just imagine them now. Please mod them down.

    The real innovation here is that instead of using a pizza ordering device to identify objects visually, we can now use a vision device to identify objects visually..

    • Re:Full Circle? (Score:4, Funny)

      by Entrope ( 68843 ) on Wednesday January 17, 2024 @07:23AM (#64166495) Homepage

      I don't think Benj Edwards at Ars knows what "comes full circle" means. Alternately, someone else chooses Benj's headlines and has trouble with English.

      Randall Monroe, in fact, foresaw [xkcd.com] this turn of events. We have come full circle back to xkcd!

      • We have come full circle back to xkcd!

        Full circle, as in returning to the situation where is hard to identify a bird in the picture?

        • Re: (Score:3)

          by Entrope ( 68843 )

          Well, that "bird" is a stick figure. She is reportedly [explainxkcd.com] nicknamed Ponytail and doesn't represent the same character in each appearance. She appears to be intentionally indistinct, so we're not even supposed to identify her.

      • Re: (Score:3)

        by mjwx ( 966435 )

        I don't think Benj Edwards at Ars knows what "comes full circle" means. Alternately, someone else chooses Benj's headlines and has trouble with English.

        Randall Monroe, in fact, foresaw [xkcd.com] this turn of events. We have come full circle back to xkcd!

        First off, I like that particular XKCD...

        However the GP was correct, to come full circle means we're back at where we began or managed to repeat the same mistake. The phrase that should have been used is "life imitates art"... As much as I deplore and abase myself for being a language pedant here (language pedants are the most odious of bores) it does show how bad journalism has become that people don't know the meaning of common phrases.

        Then again, it's Ars... Some people have delusions of grandeur,

    • Re: (Score:2)

      by EvilSS ( 557649 )
      Ah /., always focusing on the important part of the story.

    • Re:Full Circle? (Score:4, Insightful)

      by Benjedwards ( 1913440 ) on Wednesday January 17, 2024 @05:08PM (#64168471)
      Benj Edwards here, author of the piece. I wrote the headline, and you're correct that I had a misconception about what "comes full circle" means (no one else seems to have caught my error either, even in the Ars comments). I apologize for my mistake. We're a small publication (in terms of employee count), and reader feedback helps us a lot when we make mistakes like this. I think I found a better alternative and have changed the headline.

  • Not for its (unusual) lack of prescience - let's face it, I doubt most people at the time expected AI to move as quickly as it did, after such a long period of stagnation - but for its timing. It was written *right* before some of the first mind-boggling breakthroughs in AI, like Google's Deep Dream, came out. I often think about that specific comment, as a "bookend" to the modern AI era.

    • Well to be fair, Randall Munroe did write 5 years & it's taken 9. But being 4 years off in futurology is still pretty impressive.

      • Re: (Score:3)

        by aitikin ( 909209 )

        Well to be fair, Randall Munroe did write 5 years & it's taken 9. But being 4 years off in futurology is still pretty impressive.

        Well, FTA (or, at least on of the articles), "Around five years have been invested in the development and construction of the AX Visio," so...

      • Re: (Score:2)

        by jmccue ( 834797 )

        If you work on IT, the estimate is spot on :)

        When I estimate development time, I always double what I really think it would really take. Usually the amount I come up with is close.

      • I love the fact he has another comic on the phrase "In five years"

        https://xkcd.com/678/ [xkcd.com]

    • it's just that the research being done was so complicated that we didn't notice it being done until they had practical applications.

      • Neural networks have been around for decades and common student projects in CS. The vanilla NNs we used to code up in Lisp were incredibly limited. The twist was things got interesting one we added convolution and more layers. And practical implementation happened when we had enough RAM to hold the training models and enough well processed data to feed it. This was an evolution, not really a revolution. But people ignored the progress because research and hardware took 50 years to yield a useful implementat

        • Re: (Score:2)

          by Rei ( 128717 )

          Unsupervised learning was another big advance. Turns out that in order to be able to predict a dataset well where the size of the dataset dramatically exceeds the size of the weights (or one emulates such a difference with e.g. dropout or limited numbers of epochs), you have to learn all the important underlying relations about the process that generated the datasets. And when said originating process is "all of the world", well, things get interesting....

          Advancements in our understanding of and manipulati

  • Comment removed based on user account deletion

    • The first part, which is assumed to be the easiest one ("whether they are in a national park)", is way much more difficult to implement (virtually impossible, really)

      I guess my phone is virtually impossible then since it can tell me whether I'm in a national park of just about any country on the planet. Ever heard of GPS and Google Maps?

      • Comment removed based on user account deletion
        • "works best in urban areas"

          https://geospy.web.app/ [web.app]
          • Comment removed based on user account deletion

            • Re: (Score:2)

              by ceoyoyo ( 59147 )

              You can pretty much do that now. Not for any random bit of forest, but for anything that's frequently photographed it's not terribly difficult to compare chunks of a picture against large databases of other pictures.

              Post a picture or a video with a lens artifact in it on the Internet, call it a UFO, and an army of amateurs will do it for you.

              • Comment removed based on user account deletion

                • Re: (Score:2)

                  by ceoyoyo ( 59147 )

                  The image databases aren't small. Google Streetview is a hundred petabytes or so, and their image search database is probably many times bigger.

                  There are robust hash algorithms for matching images that don't require the slow pixel-to-pixel matching you seem to be imagining. Like Shazam except for images. Some of these are conventional, while others use AI models to map semantically similar images to similar descriptor values.

                  • Comment removed based on user account deletion

                    • Re: (Score:2)

                      by ceoyoyo ( 59147 )

                      I specifically mentioned that image matching can be done by conventional algorithms or AI. Sub-image matching is absolutely a thing, and can be accomplished either way. Conventionally, keypoint matching with things like SIFT or SURF, or the maximally extremal stable region algorithm are popular. The SI in SIFT stands for "scale-invariant", which these algorithms typically are. They encode descriptors from regions a couple pixels in diameter up to the scale of the whole image.

                      You can train a neural network t

                    • Comment removed based on user account deletion
      • Comment removed based on user account deletion

      • Your phone would be virtually impossible without the GPS - a very, very complex system that took decades to develop and polish. And a shot down airliner with 300 people in it to have it released to the public.

        • Comment removed based on user account deletion
        • Hardly. My phone can probably give a good indication if I am in something the size of a national park even without GPS if it can connect to a cell tower. However, that is less reliable and far less precise. Indeed most phones use this to help speed up the acquisition of a GPS signal. Plus now there are alternatives to GPS such as Galileo which is GPS-compatible but independent.

  • Oblig (Score:5, Funny)

    by stealth_finger ( 1809752 ) on Wednesday January 17, 2024 @08:26AM (#64166605)
    Oblig xk....wait
  • Really not that much more expensive than high end binoculars. ($3,000+ for Leica Geovids and Zeiss Victory's)

  • This is the obvious evolution of this seminal app.

    https://www.engadget.com/2017-... [engadget.com]

  • Actually they exist and it makes a lot more sense to put the recognition and recording in a smartphone that can be easily changed than in the source. It also means the same binocular can be used with different apps for different identification purposes.
  • Can't wait for the military applications for this technology!

  • The real question is, Can It Recognise Different Types of Trees From Quite a Long Way Away?

    For example, the larch.

Slashdot Top Deals

It is better to travel hopefully than to fly Continental.

Close