There have been many instances of companies using facial tech in the sly: shopping malls and now this. But how helpful is it really knowing most university students are the exact age and gender you would absolutely expect?
That camera hole was surprisingly crudely hacked into the case.
I’m trying to figure out why a vending machine operator would care about recognizing the faces or customers.
How will they ever know consumption of Mello Yello among Black women aged 18 to 29 is down 0.19% without massive illegal surveillance and data mining?
Several years ago, I worked in IT for a large corporation operating stores all over the country. They were considering a system where they would connect the cameras over their entrance doors to facial recognition software, and compare each person entering to a database they would maintain of every person arrested for shoplifting at any of their stores nationwide.
If they matched closely enough, store security would be alerted to ask the person to show identification. If it was not the shoplifter, they would apologize (and maybe offer a small coupon). If it was a previous shoplifter, they would be held and arrested for illegal trespass. And if they refused to identify themself, they would be asked to leave the store property.
This 2-step process was because the recognition software (and the camera quality) was not completely accurate, especially for smaller population groups. There were also questions about the cost of installing this compared to shoplifting losses. And some worry about possible bad public reaction.
I left that company, so I don’t know if it was ever actually installed widely. I haven’t heard about it.
If you’re not doing anything wrong you’ve nothing to worry sbout!
I read that Ukraine is using it to identify battlefield dead.
I’m sure it’s something like that, although is it illegal? I mean, you are in public in most situations where this would be used, and there’s nothing to tie an image in front of a vending machine back to any particular person (absent some sort of facial recognition database that ties to specific people). They would literally be using it to identify what men vs. women purchased and in what quantities, and probably something to do with skin color.
If you’ve got a vending machine, your feedback is probably pretty blunt- you just see what sold since you last refilled it. You don’t know who or when or anything like that. But if you’ve got a couple hundred machines, and you can start to recognize patterns among your customers, you might be able to get ahead of things and start pre-emptively stocking in ways that better serve your customer base at each particular machine, which theoretically at least, would result in higher sales.
For example, if you notice that during finals week, all the Mountain Dew gets bought in the first 3 days, and a lot of people come up and try to buy it, but it’s out, you might schedule a restock before it runs out. But on the old schedule, you’d just show up the next week on your regular schedule and notice that it’s out, without knowning when or who.
So if you quickly sell out of all the Mountain Dew: Smegma Surprise from many campus machines, you could probably just restock it frequently without alienating customers or breaking the law.
But you could very easily track when the machine is selling each variety, without need for the camera. If that were all that they were doing, it’d be completely uncontroversial.
If the person uses their credit card in the machine, they could tie the face to the card info. (Is that legal?)
What if the machine was linked into the school’s database of student photos?
In many jurisdictions all over the world, the use of facial-recognition technology is governed by reams of regulations, even when it is not outright illegal.
To pick a random example:
In Canada, the Office of the Privacy Commissioner of Canada and the commissioners for Alberta and British Columbia investigated the use of LFR by a shopping mall owner to monitor footfall patterns and estimate demographic information about visitors. In October 2020, the investigation concluded that the LFR processing was not within shoppers’ reasonable expectations, that there were inadequacies in the transparency measures, and that the organisation had not obtained valid consent. The commissioners recommended that the organisation either obtain “meaningful express opt-in consent” and allow individuals to use malls without having to agree to their personal data being processed, or cease using the LFR system.
I guess it depends on the type of machine, but I was thinking specifically of being able to track the events when someone walks up, looks at the machine, realizes that their choice is out of stock, and leaves. Or maybe if you’re sophisticated enough in your analysis, you can tell when they opt for their second choice.
That would be useful information if you’re trying to keep it stocked - maybe your back-end software would be able to report on that- not only have you been out of Mountain Dew for 3 days, but 37 people came by and left without buying anything during that period.
I’m not exactly sure what vending machine purposes it would serve to know that the people buying Coke are white males, and the people buying Diet Coke are Asian females, etc… The cynic in me says that it’s just an attempt to gather data that they can sell elsewhere that’s not useful for the vending machine operator, but might be for Big Soda.