It was late fall, the Texas sun sufficiently merciful, and I was standing on a wood platform overlooking the humid, green core of Hazel Bazemore Park on the western outskirts of Corpus Christi. Though forgettable by most metrics, this 87-acre swatch of honey mesquite and lowland grasses ranks as one of the country’s richest and most diverse places to see migrating raptors. Three North American flyways converge here and by the time I’d arrived in early November, more than 1.2 million broad-winged hawks, turkey vultures, Mississippi kites, kestrels, and more had passed through, a record number for the second year in a row.
The park was a fitting testing ground for some super cool tech aimed largely at birders, still one of the fastest-growing cohorts of outdoor enthusiasts. I joined about a dozen of the country’s more influential bird nerds—each of us armed with a pair of the AX Visio, the most advanced set of binoculars in the world. Swarovski Optik, the Austria-based offshoot of the larger, highly secretive, luxury crystal brand, Swarovski, would not introduce them to the world until Jan. 9, 2024, and as such we were asked not to let other birders see them up close. That’s because the AX Visio can do what no other binocular or scope in the world on the consumer market today can do: tell you what you’re looking at, instantly, at least when it comes to virtually every known bird on the planet. The binos can also identify hundreds of species of mammals, and soon, butterflies and dragonflies. And that’s just the start.
“If you have a database on wildflowers or mushrooms or stars or whatever, we can train the system to identify them,” Ben Lizdas, Swarovski Optik’s business development manager. “The idea is absolutely for developers to be able to contribute to this. It’s limitless.”
The identification feat alone is groundbreaking, but so are the other tasks that the smart binoculars can handle. The AX Visio has a tagging feature that allows you to drop a pin on a certain subject or location, like a mountain goat on a distant ridge. Hand the glass to a friend and a reticle in your field of vision will direct the viewer to the exact spot you just pinned. Paired with a smartphone app, the binoculars can also stream a live feed to up to four other devices at a time (though they need to be within about ten feet of each other), so everyone on safari can watch the jaguar eat. The display projected directly into the viewfinder can also show a digital compass offering both cardinal directions and azimuth angles. It has an onboard camera, a GPS, and Bluetooth capabilities for firmware updates and app connectivity. To function fully, it needs no signal from anything whatsoever.
Of course, Swarovski optics aren’t cheap, think $3,000 or more, and neither is the AX Visio. With an MSRP at about $5,330, they are not for everyone. But after three days of playing with them in Texas, followed by another two weeks around my home in Oregon, I can say they’re a blast, intuitive, and offer all sorts of implications for birders, hunters, guides, travelers, and those who just like to navigate the natural world by name. “I think there are applications that we can’t even imagine yet,” says Janet Moler, a manager with Portland (Oregon) Audubon. “I don’t recall anything on the market even close to this.”
At the moment, in Texas, I trained the glass on a flamboyant yellow bird with a sky blue head and a black throat that had materialized as if out of a Walt Disney film. I’d never seen a bird like it. The words “Green Jay” illuminated in a simple, unobtrusive orange font almost instantly along the bottom of the hyper crisp image. Suddenly I could appreciate this park a little more.
Swarovski’s AX Visio binoculars (Photo: Courtesy Swarovsky)
The Swarovski AX Visio Binoculars Basics
The AX Visio (a riff on something like “augmented-experience vision”) looks like a chunkier, more militaristic set of binoculars with 10×32 lenses, which, like all optics, translates to “how big by how bright;” In the AX Visio’s case, 10x magnification with 32 millimeter objective lenses. That’s enough glass to collect sufficient light for most outdoorsy applications, though a little under-gunned for dimmer conditions, like in the thick of a rainforest.
Like all of Swarovski’s optics, the lenses are crafted to nanometer precision. But the real magic lies with a fiercely guarded mix of light-altering chemical coatings applied in as many as 50 layers to enhance clarity and contrast. So secret is the recipe for these coatings that Swarovski has opted not to patent them, a protection that would require the company to divulge its materials and methods. Better to let the competition spend the time and money trying to reverse engineer it all with lasers and gas spectrometers than to spell it out in a patent, or so the thinking goes.
Some of the experts on this trip, most of whom are on the front line of the birding world as dealers who sell optics to other birders, said they could see birders wanting a brighter lens, say 42 mm, but the smaller 32 mm glass saves on weight. That was necessary because a third optical barrel sits under the bridge linking the other two barrels near the focus wheel. This third barrel houses a 13 megapixel camera with a fixed 2.2 f-stop (and maximum exposure time of 1/125) that’s capable of shooting HD video at 30 to 60 frames per second. Figuring out how to pair a camera with the binoculars was key to the whole “smart” process.
“People have tried to put cameras in binoculars before, but the technology was changing so fast that by the time it came out it was already obsolete,” says Daniel Nindl, the company’s head of product management. Now the components are so small and the processors so robust that Swarovski Optik, which first began toying with smart binocular designs about a decade ago, felt confident enough to move into prototype stages about six years ago. In the end, engineers packed the AX Visio with 37 lenses, eight prisms, and nine electronic boards powered by a removable, rechargeable lithium battery pack. Combined with the 390 individual parts (think sensors, a gyroscope, magnetometer, accelerometer and more) all housed in a forest-green, IP68-rated moisture- and dust-tight body, the unit is about three pounds. Though it’s a bit of a beast, it feels great in your hands.
How Does The AI-Generated Identification Work?
The ID feature is well worth the heavy weight, especially when you consider some of the alternatives. I got into birding the way many others did during the pandemic, when I happened to look out my window to find a delightful little guy with a body as yellow as an Easter Peep sporting a jaunty black patch atop his tiny birdy crown. I went full analog and hauled out a dusty copy of Birds of North America, eventually landing on page 825, a Wilson’s warbler.
Though I’m not a serious birder, I’ve learned a few very basic tricks that make identifying birds so much easier than that, namely the free Merlin Bird ID app. Merlin is a “machine learning -powered bird ID tool” put out by the avian gods themselves at the Cornell Lab of Ornithology at Cornell University. I take a picture of a mystery bird with my smartphone, record its song or enter its description, and Merlin will spit out a list of species possibilities. I use it almost every day just walking the dog.
This app, or rather the research and database behind it, forms the muscle behind the AX Visio’s ability to identify birds. Swarovski Optik obtained the rights to use the data and then figured out how to pack the hardware and software into binoculars that could make sense of it. The system includes a processor similar to one in your phone that powers an algorithm using a “neural processing unit” based on the Merlin app. In short, artificial intelligence.
What’s truly amazing is the sheer size of the bird reference library that the AX Visio’s AI can tap. On your phone, the Merlin app is so data-heavy with photos and sound bites that you have to pick and choose which “bird packs” to install based on your geographic location. Those packs narrow down the realm of possible birds to those the user is most likely to see in that particular area. For example, that’s 717 birds for “U.S. and Canada: Continental” and 810 birds for “Costa Rica.” The AX Visio, meanwhile, can only identify birds by “sight.” That data set consumes less storage space than the full Merlin app and allows the binoculars to have an on-board reference library that essentially includes every bird from every bird pack—roughly 8,000 birds total. The processor needs no connectivity to access it. That means it can identify a satyr tragopan in a remote rhododendron forest in Bhutan as easily as a mourning dove in New York City.
Understanding how this all unfolds is material for a graduate degree, but as a user, it couldn’t be more simple. To identify the green jay, I powered the unit on, waited for it to lock in a GPS signal, and turned the “mode wheel” to the bird ID function. Looking through the viewfinder, which included adjustable eyecups I could dial in to fit my sunglasses, a reticle appeared in the shape of a circle cut into four equal, curved segments. The more the bird can fill that circle, the higher the confidence in the AI result. Holding the binoculars naturally, you can toggle the size of the circle by depressing a button using a right forefinger.
A second, adjacent button acts like a shutter release on a camera. Once I had the jay lined up in my sights, I depressed the second button halfway, and the unit’s autofocus took over. Depressing it the rest of the way, the unit took a picture that was automatically uploaded to my paired iPhone 12 and then produced an ID readable directly in the viewfinder. All four segments of the circle had grown thicker, like a font in bold, signaling a high-confidence result. The fewer segments in bold: the lower the AI’s confidence in its result.
The system isn’t perfect. The unit couldn’t decide if a seagull was a laughing gull or a Franklin’s gull—it kept alternating between the two each time I hit the release. Ideally, the viewfinder needs to be filled with 224 x 224 pixels, though the minimum it needs is 100 x 100. Other times, it tested my own limitations. A hummingbird showed up and it was difficult to get a clear, stable shot of it that filled the circle. I tried anyway and the unit said it couldn’t recognize it or it wasn’t in focus. When I did get a decent shot of the bird face-on, the AI thought it was an Allen’s hummingbird when the birders all knew it was actually a Rufous. The differences between the two are difficult for beginners to distinguish and boil down to the shape of the tail feathers.
“To be fair,” said Clay Taylor, a naturalist who joined Swarovski Optik in 1999 as the division’s first in-house bird specialist, “even (famed ornithologist) David Sibley would need to see its back before he could tell you what it is.”
Other times I was shocked the AI could make sense of what I fed it at all. A raptor rocketed by and I fired off a sloppy shot. The unit called it a northern harrier, correctly. In the most comical, extreme example, perhaps, a bird never seen anywhere north of Panama suddenly showed up in downtown Corpus Christi, having likely hitched a ride on one of the many ships that come into port—a vagrant in birder-speak. The AI must have overridden what its own GPS said it couldn’t possibly be (or perhaps the unit couldn’t lock in a solid GPS signal among the buildings) and identified it as a cattle tyrant, a fly-eating fiend with a yellow breast and olive-brown back that had found an endless feast in a blue downtown dumpster. Soon hundreds of people from all over the country had gathered around this greasy trashcan to catch a glimpse.
Additional Features
I played with the other functions over the next few weeks. The location tagging setting led me right to a vermillion flycatcher a friend had spotted on a fence. Another time, I put it on mammal mode and drew a bead on a shih tzu walking down the beach, but all it could say was “dog.” One day, for giggles, I pointed it at my teenage daughter. “Human,” it said, though the state of her habitat would suggest otherwise.
I wondered if hardcore birders who already know hundreds of species by heart would have a need for something like this, and the answer is maybe. “We can all go someplace new and be totally lost,” says Diane Porter, co-founder of birdwatching.com. Moler of Portland Audubon agrees: “There were many birds in Texas it identified when I didn’t have a clue,” she says. Having a pair to share among a group seems ideal, like an outfitter that equips its trip leaders and guides with a set to help clients see an owlet in the redwoods or a lion snoozing in the shade.
I shot a video of one egret bullying another in a pond and took pictures of distant buildings with architectural features I thought were cool, which was easy since I didn’t have to put down the binoculars and take out my phone or a camera. Though I enjoyed the photo function, it still can’t compete against a dedicated camera with, say, a 600 millimeter lens. “No way will it replace my big camera,” Porter says.
The most exciting features, however, may be the ones to come. The mode wheel already includes two empty slots ready to be claimed by future functions (which will also be open to third party developers). Perhaps one day one will go to the names and elevations of distant peaks or even climbing routes to the top. Maybe plane spotters will find a way to use it. In the meantime, there’s no doubt they’re game changers for many.
Personally, I’d kill to own a pair for the sheer amount of joy I got out of the Genesis-like gift of being able to give names to these delicate, gorgeous marvels of the world, and for the way that knowledge enriched my time in Texas, expanding doors and introducing me to others. (Like those 1.2 million raptors spotted at a tiny park for starters.)
“It’s a revolutionary product,” Moler says. “Now when a person walks in and says, for that price, those binoculars should identify the bird for you, I can say, they do.”
The post Swarovski’s AI-Powered Binoculars Tell You What You’re Looking At appeared first on Outside Online.
Tools & Tech, Binoculars and Telescopes, Birding, Hunting, Technology
