Over the past couple of years, Amazon has been slowly launching Amazon Go stores around the US and just recently opened an Amazon Fresh store in the UK. Even though their online website is the go-to place for just about anything, they fall short when it comes to selling groceries and fresh food. These physical stores are the company’s attempt at expanding into these markets, but with a unique twist- Amazon is creating smart grocery stores engineered with security cameras and artificial intelligence.
“Just Walk Out” Technology
From the outside, the Amazon stores may look like any other ordinary grocery or convenience store, but the inside couldn’t be more different. Unlike traditional stores, there are no cashiers, no lines, and no checkouts. These stores were developed with what Amazon calls the “Just Walk Out” technology. This enables customers to just walk in, pick out their things, and walk right out. Instead of having to wait in a line to pay with a cashier, the customer will be automatically billed through their Amazon account. But how does the system know which products to charge you for? Amazon is able to do this by using artificial intelligence.
Amazon’s Artificial Intelligence: How It Works
Each Amazon store utilizes a system of artificial intelligence, cameras, and sensors in order to create a smart shopping experience. Upon entering, each customer will approach a turnstile to scan a QR code from the Amazon app that is linked to their account. Once they start shopping around the store, artificial intelligence comes into play.
For Amazon’s business model to succeed, they must first accurately identify each unique customer. If they cannot differentiate each customer, then they risk charging people for another person’s order. Amazon is able to identify different customers by using security cameras programmed with artificial intelligence learning algorithms. One of the algorithms that they use is object recognition; this is an algorithm that is able to identify different objects and their location. But how does this tie into person identification? With object recognition, the system is able to detect thousands of different objects. This means that it is capable of recognizing each person as being different from one another. However, identifying each person is just half of the problem. Each person’s movement also needs to be monitored in order to track which items they are picking up. Object recognition is used to detect different customers and track each person’s path throughout the store to ensure that the orders don’t get mixed up.
As mentioned before, object recognition is an algorithm that differentiates objects. Besides being used for person identification, Amazon uses this artificial intelligence learning algorithm for item identification. Item identification is important for the system to know which products they should charge each customer for. Like person identification, the security cameras are capable of detecting different items and their movement. If a product is detected to be placed into a cart and taken out of the store, then the system will know specifically which product it was and what to charge on the account. The algorithm is also capable of recognizing when an item was placed back on the shelf to prevent charging the customer.
Another artificial intelligence that Amazon has mentioned that they use is face recognition. Face recognition allows the security cameras to detect and recognize faces in real-time. With face recognition, Amazon is able to pull up customer profiles. These profiles contain customer information such as their gender, age, and purchase history. This information helps the system to be more accurate with item identification. Even though object recognition works most of the time, it’s not always capable of differentiating specific items, especially if two items look very similar. Pulling up the purchase history helps the system determine which item was picked up based on the customer’s past preferences. Additionally, the profiles help Amazon to create targeted ads and recognize repeat customers to offer them in-store deals.
Using a combination of security cameras and artificial intelligence algorithms, or video analytics, is not a new concept. Video analytics have been used by many retailers in order to gain business insights and improve security. However, Amazon has taken this concept a step further and adapted artificial intelligence to automate the shopping experience. It’ll be interesting to see what other things security cameras and AI will be used for in the future.