Amazon Go, the e-commerce giant’s automated checkout project, is just preparing to open its second convenience store. But like the Soviets’ launch of tiny Sputnik in 1957, Amazon Go’s 2016 debut is already fueling a (retail) space race. The latest company to launch is Israel-based Trigo Vision, which officially emerges from stealth today with $7 million in funding.
Like other potential Amazon Go challengers, Trigo Vision uses computer vision to recognize what each product looks like when it goes into a customer’s hand or shopping basket. No human has to ring up individual items. The systems can also, of course, keep track of individual customers to know who took what, though not necessarily by matching them to their actual identity. Shoppers can allow themselves to be identified when they enter the store, or they can anonymously pay the bill that’s presented when they reach checkout–according to Trigo Vision.
Amazon is focusing on fairly small stores: Its new one measures about 3,000 square feet. And the company claims to not have plans to expand to its Whole Foods sites, which average 40,000 square feet. Trigo Vision, however, says that it can scale up to full-size grocery store dimensions. “We’re already working in some of the biggest retailers,” says Michael Gabay, the startup’s CEO.
Gabay doesn’t quite say how the company can handle such sites, other than to brag about having a top-notch computer vision team. One hint, though, is that it focuses its training–teaching the system what each product looks like–on real-world settings, with the kind of low-quality images that basic IP cameras in stores capture. (Gabay says they need about 50 cameras per 200 square meters, about 2,200 square feet.) “We don’t want our data to be too clean,” says Gabay. “We want to train our system on real places.”
Where are those places? Trigo declines to say, other than that it is testing the tech in a single European store–and it is in discussions with others. That’s a fairly common response to what other auto-checkout startups like AiFi, Standard Cognition, and Aipoly say. Meanwhile Microsoft–a potential giant rival–hasn’t even announced a program yet, though reports of its efforts have leaked out. Trigo Vision projects that it can launch an actual service in, “a year, more or less,” says Jenya Beilin, the company’s COO.
Also like rivals, Trigo Vision isn’t tying success only to the lofty goal of a perfect automated checkout. “The platform is more than the ability to do the checkout approach,” says Gabay. It can handle other services, such as tracking inventory, spotting shoplifters, and seeing what placement of products (such as promotional end caps) attract the most takers. Customers using an app could even search for a product and get directions to it.
“We have the ability to track everything in the store,” says Gabay. “We know where are the products, the people, the baskets, the carts, everything.”
Correction: A previous version of this article stated incorrectly that Gabay and Beilin are brothers.