How To Identify A Picture is a question that Wolfram comes with the perfect answer, an image searching website. Check out this post to find out more.
How To Identify A Picture In the Next Normal With Wolfram?
One of the most remarkable things you can do with Wolfram Alpha asked it what planes are flying above. If you’re on your phone, it’ll retrieve your position and cross-reference it with a database of planes, including altitude, angle, and even flight number and aircraft type.
But, in many respects, Stephen Wolfram’s latest search tool is more amazing. It intends to recognize anything in a photograph. You upload a photo and receive a computer-generated guess within a few seconds.
“IT WILL NOT ALWAYS GET IT RIGHT.”
“It won’t always get it right, but most of the time, I think it works quite well,” Wolfram writes. “And what’s intriguing to me is that when it does get something wrong, the mistakes it makes generally look incredibly human.” Based on some preliminary testing, that’s a very accurate judgment. Just type in Yosemite National Park’s Half Dome and was told it was “elevation,” while a snap of a gecko classify as a “night lizard.”
Surprisingly, it ranksd an image of a cow as “black Angus” and two bowls of ice cream as “frozen yogurt.” That’s close.
It’s intriguing to see how all of this progresses beyond bombarding a website with images of your recent vacation or what’s in your kitchen.
Wolfram believes the research beneficial if applied to massive sets of photographs in an attempt to identify and categorize them. Others can utilize the technology to integrate picture recognition into their apps. Consider the visual recognition present not only in Google+ images but also in other photo applications and sites.
TEST ON CATS, SLOTHS, AND CHEWBACCA.
Wolfram claims that the system was fed “a few tens of millions” of pictures for it to understand what was what. That “seemed pretty close to the number of unique views of objects that individuals acquire in their first few years of existence,” he noted.
The system did show difficult pictures such as cats in spacesuits, sloths in party hats, and even Chewbacca. Also, all of which fails to recognize properly, but gracefully:
It can now recognize around 10,000 common types of items. Wolfram says that it still has difficulties distinguishing individual persons, art, and things that are not “actual daily stuff.”
The new picture project joins Google’s Goggles and Amazon’s Firefly as fast identification tools. Still, it is notable for not attempting to sell you anything based on its discovery.
It also comes shortly after Flickr’s new Magic View and Microsoft’s research site that predicts people’s genders and ages based on photographs. Unlike Microsoft, Wolfram claims to preserve a thumbnail version of the photo after you’ve submitted it.
So it shares with others, and that it’s gathering photos to keep training its algorithm.