Smile, You’re on Camera. How Edge Computing Will Support Machine Vision in Stores

Ever get the feeling you’re being watched? You are. Cameras are everywhere these days – traffic lights, front doors, supermarkets, you name it. But cameras aren’t just for surveillance and security anymore. Pioneering retailers have started using them to expedite in-store shopping.

This use of cameras is called machine vision (also referred to as computer vision), with the primary goal of shortening – or outright eliminating – checkout lines. It requires real-time processing and analytics for massive amounts of data and, as such, can benefit from edge computing networks that minimize data latency and improve bandwidth.

Machine vision consists of deploying cameras and sensors tied to artificial intelligence (AI) systems to track the movement of shoppers through a store as they pick items off shelves. As shoppers push their carts through the store, the AI-powered cameras send data on all the items to a tracking system that tallies everything and performs a virtual checkout at the end. As a result, everything is done electronically, so customers don’t have to carry cash or wait at a checkout line.

The technology is already in place at Amazon Go stores to enable the “Just Walk Out” shopping model. Walmart, too, has started using machine vision at more than 1,000 stores but with a more limited purpose: to monitor checkouts and deter theft. Adoption of this technology may pick up steam as retailers look to improve checkouts and address challenges such as the current labor shortage.

How edge computing supports machine vision

Machine vision is a cutting-edge approach that requires another type of edge – edge computing. Without processing, analytics, and storage at the edge, retailers will be hard-pressed to implement such a data-intensive technology. Keeping track of dozens of shoppers as they wind their way through a supermarket, for instance, requires thousands of real-time computations.

If the data necessary to process all of a shopper’s movements had to travel to an off-site cloud, the potential for latency increases. However, deploying edge computing sites at or near the store helps to eliminate the latency issue. The data is captured, compiled, and analyzed quickly so that the checkout process can be completed as a shopper leaves the store. In addition, an electronic invoice is issued, and confirmation of a deduction from the user’s bank account or credit card is sent.

Edge computing supports machine vision and a variety of other technologies as retailers implement omnichannel strategies to create a seamless experience for shoppers who go back and forth between online, mobile, and in-person shopping. Edge technology also can play a crucial role in accommodating shoppers’ changing habits, including curbside pickup and BOPIS (buy online, pick up in-store).

Gradual adoption

Smart technologies such as machine vision have been anticipated for years, and now they are really starting to make inroads with retailers. Transactions that rely on smart technologies will reach $387 billion in 2025, a massive jump from $2 billion in 2020, according to Juniper Research.

Of course, complex technologies like machine vision will require massive investments, so adoption will be gradual. Retailers probably will deploy it for a specific function first and then expand use cases as budgets allow. For example, one area where machine vision can help expedite checkouts is through the use of cameras to recognize produce, so shoppers don’t have to navigate through multiple screens at self-checkouts to find PLU codes for apples, oranges, and lettuce.

No matter how fast or slow the adoption of machine vision is, one thing is certain – edge computing will play an integral role in the real-time collection and analysis of the massive amounts of data required to make it all work. As such, edge computing will be fundamental to enhancing the retail customer’s experience for the foreseeable future. Learn more about edge computing and its role in digital transformation.

Leave a Reply

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