Smart vending machines and self-serve kiosks are more prevalent in the world now than ever before. Everything from fresh lattes to T-shirts to new headphones for your upcoming flight can easily be purchased with the swipe of a credit card. Smart vending machines have made it easier for consumers to purchase higher priced items while allowing vendors to save on labor costs. However, age restricted products like alcohol and cigarettes remain a hurdle for self-serve purchasing with nobody available to check ID. Advances in age assurance technologies, in particular AI-driven age estimation, can change that.
When purchasing in person, if someone that looks young tries to buy an age-restricted product, the clerk would simply ask to check their ID. If they are old enough, then they buy the product. If they are not old enough, then the clerk rejects them, and no purchase is made. In either case, no customer information is saved, and data privacy is not an issue.
In this AIoT world we now live in, AI-driven age estimation technology can operate as that clerk in the store, providing assurance the customer is of age. Age estimation software uses AI computer vision and machine learning to estimate a person’s age based on their facial features. And just like with an in-person purchase, no customer information is stored on file. The customer’s information remains private and is not sold off to marketers.
Here is just a small handful of applications for age estimation technology:
A typical bottleneck when using self-checkout kiosks in supermarkets is purchasing alcohol. You scan all your items, set them in the basket, but that bottle of wine requires waiting for an attendant to come, verify your age and ID, and then you can continue. Or you simply have to avoid the self-checkout kiosk entirely. AI-driven age estimation technology can be added into the kiosk and its cameras scan your face to verify legal age and continue the transaction.
Adding age estimation to vending machines enables the sale of age restricted products for improved efficiency. Bars can sell cigarettes with age estimating vending machines and allow the bartenders to focus on making cocktails. Pro sporting events could sell beer at vending machines, freeing up concession lines to only focus on food orders.
Shopping malls and department stores can adjust digital signage promotions based on the age demographics of who is shopping at that time. Switching from promotions targeting the elderly to young adults is effortless and can be adjusted based on the shopper analytics gathered by age estimation software.
Under the U.S. government’s Department of Commerce, the National Institute of Standards and Technology (NIST) conducts a Face Analysis Technology Evaluation (FATE) for Age Estimation and Verification (AEV), an ongoing evaluation of software algorithms that inspects facial photos to produce an age estimate. In their latest findings released on April 14, 2025, testing was conducted for age estimation of people aged 18 to 30, using 2.7 million images. 23 different age estimation algorithms were submitted, with an average of 3.845 mean absolute error (MAE). The MAE is the range of variance above or below the person’s actual age, so a lower value MAE indicates better accuracy.
FaceMe® has excelled in previous NIST evaluations, including FATE Presentation Attack Detection (PAD), Facial Recognition Technology Evaluation (FRTE) 1:1 Verification, and FRTE 1: N Identification. Now FaceMe® is leading the way in emerging age assurance applications. FaceMe® tested remarkably well in NIST’s Age Estimation and Verification evaluation, providing very encouraging results:
FaceMe® performed above the competition, delivering high marks in age estimation testing. With a MAE score of less than 3.282 years, FaceMe® ranked among the top 10 algorithms tested.
FaceMe® performed exceptionally well with all image types tested, including visa photos, border crossing images, and mugshots, demonstrating FaceMe®’s suitability to be used worldwide for a multitude of applications.
There has been ongoing discussion regarding the effects of social media on minors. Australia is the first country to ban children under 16 years of age from social media, with other countries such as Norway, France, Germany, and Italy considering similar age-based restrictions or requiring parental consent for minors under 15. Germany is the first to approve the use of AI age estimation software for age restricted online content access with a 5-year buffer, requiring users to appear at least 23 years old to access content meant for those 18 and up. Many more countries have proposed legislation for age assurance measures for the protection of minors online, and it could only be a matter of time before these proposals become law.
The future of AI age estimation technology looks very bright. It enables self-checkout of age restricted products in an efficient manner while remaining non-intrusive to customer privacy concerns. As technology matures further and MAE decreases, it’s likely more and more governments will approve the use of age estimation technology, and its applications will continue to grow.
