AI Inspiration #12: Funniest Computer Vision Fails; Shazam for Fashion; Aibo Returns
Here’s everything that’s new in artificial intelligence and computer vision, with a little tech pop culture to make the medicine go down. Our logic is undeniable.
Is that a turtle or a rifle?
8 Times Computer Vision Hilariously Failed
You heard the one about Google’s AI mistaking a turtle for a rifle? That’s just the latest computer vision blooper — a common occurrence with any emerging technology — but here are eight others that definitely include everything from Apple and Samsung to amusingly oddball AI-created smartphone cases and self-destructive robots.
Shazaming the Kim Kardashian Look
Celebs on Instagram may spur sales or aspirational envy when they share and tag images and posts of clothing and accessories, but all too often the stuff is out of reach for the average fashion-focused fan. So kudos to Kim Kardashian for advising a ScreenShop, a new smartphone app that lets users upload images they capture or find online of clothing or accessories they like, then uses computer vision to find similar items in multiple price ranges from dozens of retailers including H&M, Nordstrom, and TopShop.
Designer Robots
The future of creativity involves humans working with AI, but how? Here’s one idea: AirBnB has designed a machine-learning system that recognizes low-fi wireframes, the symbols designers used to initially sketch out user interfaces, and automatically generates them into code. This is great for UI designers, but whither the creative coders of the future?
Did Somebody Order an Assassin?
Did Amazon jump the tech shark with its new computer vision-enabled Key service? Great satire is often the first sign. Regardless, The Onion offers this hilarious take on the delivery service of the future, which might make for a biting horror comedy movie one day.
Out of the Doghouse
After a decade’s absence, Sony’s ahead-of-its-time Aibo robo-dog is returning to the market, in Japan, at least, this time with more expressive OLED eyes, improved computer vision that learns from pictures it takes, and a slew of competitors such as Sharp’s Robohon and Mitsubishi’s Pepper.
Automating the Automation
Good AI talent is scarce, but the need for machine learning is ever-growing. So, the only way to design and build more AI is to create AI that can design and build AI on its own, at least that’s what’s behind initiatives such as Google’s OpenML. Welcome to Neural-Network-Building-as-a-Service.
Anyone you know interested in computer vision? Forward this to them so they can subscribe, too. And please submit any computer vision stories you think we’d be interested in posting.
The Visionary newsletter is produced by GumGum.