Amazon’s Just Walk Out technology has received an upgrade with advanced artificial intelligence technology that promises to significantly improve accuracy, help reduce acquisition delays and increase the ease of delivery of technology for sellers.
In a blog post, Jon Jenkins, Just Walk Out’s Vice President of Technology at AWS Applications, said Just Walk Out tech started in 2018 with the latest AI and machine learning available at the time to find: “Who took it?”
The system analyzed the customers’ behavior “in sequence” – their movement, their location, what they took and the quantity of each item – with each action being “used individually. “
Jenkins added: “However, in unusual or novel situations (such as when the camera’s view is obscured by poor lighting or a nearby customer), the tracking process may take time to know what you bought with confidence, and sometimes you need to practice again by hand.”
The new advanced AI-driven system analyzes all sensor data “simultaneously, rather than sequentially,” Jenkins said. Multiple cameras, weight sensors, and other information help the system to prioritize “what is most important to accurately identify the types and quantities of selected items.”
He adds, “For example, a customer may pick up and put away several types of yogurt, with different ingredients, and while they are doing that, another customer may find the same thing, or the refrigerator door can collapse, obscuring the view of the cameras. In complex situations like this, the new model can quickly and accurately identify the actual items taken by each customer.”
The new multivariate model increases accuracy by using the same machine learning models based on transformers based on multiple GenAI tools online and in physical stores. Transformers are a type of neural network architecture that processes sensor data into outputs, such as shopping receipts without payment.
News of the update confirmed Amazon’s commitment to expanding Just Walk Out after an April report by The Information that the technology was being removed from Amazon stores led to questions about its effectiveness.
Jenkins said Just Walk Out is available in more than 170 third-party locations, including airports, stadiums, universities and hospitals, in the US, UK, Australia and Canada. Plans are to more than double the number of other stores using the technology this year. Some New UK stores and Amazon Go are using the technology. Jenkins said, “As we scale, the system will continue to learn from day-to-day retail situations and raise the bar for accuracy and convenience, delivering AI benefits to both retailers and customers.” around the world.”
The Information Report found that despite its first appearance as a completely independent solution, it is very dependent on human reviewers, often leading to late customer visits. Critics of Just Walk Out’s technology in the past have questioned the cost and structural challenges associated with installing and maintaining video cameras and other equipment on store ceilings.
In a follow-up interview at the time with Sucharita Kodali, VP and principal analyst at Forrester, Jenkins attributed Fresh’s turnaround to strong customer demand for its smart Dash Cart. He said The Information’s finding that 1,000 people in India watched and wrote videos to write ads by hand “has no basis in fact” but noted that the technology works better for smaller brands, especially where planograms that were too similar and too small for a person. intervention is required.
Kodali wrote, “This latest news reinforces the concern people have had all along about making technology cost-effective in large, low-cost areas like grocery stores. .It seems to thrive in high-end, small-scale settings.”
#Advanced #Upend #Amazons #Walk #Technology #RetailWire