Google Cloud launches shelf inventory AI tool for retailers

admin

imageGoogle cloud It says it has developed a new artificial intelligence tool designed to help big box retailers better track inventory on their shelves, and aims to improve on technology that has struggled to work well in the past.

Google Cloud said Friday that its algorithms recognize the inventory of consumer goods products on shelves from videos and images provided by the retailer’s own ceiling-mounted cameras, self-driving robots with cameras, or store associates.You said you can analyze it.The tool, which is currently in preview, will be broadly available in the coming months, the company said.

alphabet of the corporation

cloud business announced technology, Along with a suite of artificial intelligence tools aimed at e-commerce ahead of the National Retail Federation conference in New York City.

The lack of timely and accurate information on shelf inventory is a major problem for retailers and so difficult to manage that mere guessing has become the industry standard, said a retail deputy at an IT research and consulting firm.President Robert Hetu said.

Gartner According to Carrie Tharp, Google Cloud’s vice president of retail and business, that information helps retailers queue in different ways, such as replenishing out-of-stock items faster and reducing lost sales opportunities.can be filled.consumer.

“If every retailer knew what was in store and how much was left on the shelves, their lives would be a lot easier,” she said.

The idea of computer vision-enabled shelf-checking technology has been around for a few years, but during that time it didn’t quite work.Hetu said retailers are hesitant due to the cost and complexity of large-scale camera deployments.

Data is also a problem, according to Tharp.

Retailers historically have not had access to complete, organized and labeled data on all product offerings.

According to Google Cloud, building the AI model itself requires understanding how to perceive products from different angles, under different lighting, and in imperfect real-world conditions, such as seasonal packaging changes.there is.

Google Cloud says its products are trained on a database of over 1 billion products, including images licensed and published directly by manufacturers.Its algorithms are also designed to recognize those products.The human eye sees a box of cereal whether the image is from a ceiling-mounted camera or cell phone video, whether viewed from above or from the front.Google Cloud the same way you do Said.But there are still challenges.

“It’s probably not fully solved yet,” said Graham Watkins, executive vice president of supply chain transformation and retail innovation at supermarket chains.

Giant Eagle Co., Ltd.

He said early tests in the Giant Eagle Innovation Lab, designed to replicate store conditions, have shown Google Cloud products to be more than 90% accurate.That’s enough to attract continued interest from supermarket chains, but not yet enough for the company to consider a large-scale rollout.

Currently, Giant Eagle offers He said he continuously gives feedback to Google Cloud about where the tool isn’t working so it can fine-tune it.For example, if the camera is too high or too low and the algorithm can’t identify the product, Giant Eagle provides that image to Google Cloud so it can train the algorithm to recognize that angle next time, says Watkins.says Mr.

The supermarket chain said it plans to start piloting the technology in physical stores in the coming months, but if the company does decide to pursue broader deployment, it could roll out to the entire chain.deployment will take several years to materialize, he said.Watkins said.

Part of the reason, he said, is the high degree of associated costs.

According to Gartner’s Hetu, camera visualization of every shelf in every aisle, whether from overhead cameras or from inventory robots roaming the store, is complex and costly.Proposal.

especially

Efforts to use roving robots ended That’s because various, sometimes simpler, solutions have proven equally useful in store aisles for tracking inventory in 2020, according to multiple people familiar with the matter.

Hetu said he expects investment in shelf-checking technology to continue despite cost barriers due to the growing need to digitize the in-store experience.

But it won’t happen overnight.He says it could take three to six years for computer vision shelf checking to become mainstream.

giant eagle Watkins said the algorithm may never be perfect, and there will always be unique conditions under which it won’t work.

“There will be give and take between technology and operational business processes.

Whenever you’re in a new space, you’re always balancing.

How far is enough?” he said.“It’s kind of an iterative process.”

Write to Isabelle Bousquette at

[[email protected]](/cdn-cgi/l/email-protection)

Copyright ©2022 Dow Jones & Company, Inc.All rights reserved.87990cbe856818d5eddac44c7b1cdeb8

#Google #Cloud #launches #shelf #inventory #tool #retailers

Credit.

Leave a Reply

Your email address will not be published. Required fields are marked *

Next Post

Delta decided against using ‘fairly old’ backup system during FAA outage

New York CNN — Delta Air Lines has a backup for the federal aviation safety system that failed this week, but the airline decided not to use it, CEO Ed Bastian said Friday. Bastian described the Delta’s secondary system as “fairly old backup technology” with the notices known as NOTAMs […]

Subscribe US Now