Yup it’s for “advertising” say for example the Army wants to know which areas have the most fighting aged men. So posters and recruiters know where to hang out. (this is the most extreme example.)
Vending machine company sells facial recognition and temporal location data to a data broker who enriches it to enhance identifiability and this data is sold to a stalker who uses it to murder people. That’s a more extreme, but certainly not most extreme, example.
Because people are dumb. If the machine knows when someone is looking at it, it can stop doing whatever it does to try and get your attention, and put itself in “sales mode”.
Still, you’re right. It seems like an overly complicated and expensive solution. Old-fashioned vending machines did the job just fine.
Stanley sounded alarm after consulting Invenda sales brochures that promised "the machines are capable of sending estimated ages and genders" of every person who used the machines without ever requesting consent.
I saw some posts about a similar technology in the meetings and events industry: a company is selling “facial analysis” not “facial recognition.” They try to get around privacy laws by saying “well our technology does scan every single face it sees, but it doesn’t store that image, it just determines age, gender, race and emotional sentiment and adjusts tallies for those categories in a database.”
It’s still information gathering I didn’t consent to while attending a conference, and it’s a camera with the potential to be hacked.
Of course it’s always about marketing and advertising. They want to have a heat map of which areas are popular and at what times. In the case of events so they can sell to sponsors and exhibitors. In this university it’s less clear. Do the vending machines have a space to sell ads? That would be my guess.
Sure that’s their claim but they’re not asking ‘why have that type of tech anyways’.
If it’s supposed to just act as a motion sensor, we’ve had those for decades. None of which needed to register if it was a face or not. Why isn’t the purchasing interface just always there, why is it an interface, and why is it not just a button that says press to start…
Why is there a computer in there that’s been trained on how to recognize what a face is in order to open up a purchasing interface. What would be the point of investing that much research and development if it was just doing something that could have been accomplished in the '90s with tech that you could have bought it radio shack.
From the side of someone who works heavily in data analysis and application databases I can tell you it would be very, very easy to see if it was just a front end application using the data or storing it in a database. There are use cases for both setups, absolutely, but a cursory examination of the machine in question would make it abundantly clear which it was doing.
Article says “the machines are capable of sending estimated ages and genders” so it’s not recognizing individuals, but perhaps adjusting the sales pitch for who it sees walking by.
(But it’s a collage campus, so most students will be around the same age. Maybe it pitches different things to teachers?)
I’d doubt it’s collecting or transmitting much. It’s probably just estimating age, sex, race etc. and using it to decide which promotion to put on screen. It’s possibly collecting these to determine what type of people use the machine. Similar to those billboards in shopping centres.
Storing each individual to recognize later or identify online seems like a stretch.
If it did have a user bio database, it would be centralised and not on the machine itself.
I think the problem is that it is storing the user faces, at all. If it were simple identifying each person’s characteristics there would be no reason to save that data for later. Also, apparently the company advertises that the machine does transmit this data for estimating age and gender for every purchase.
That’s your claim though. They are storing “male, 24” and that’s it, no face. Of course they could be lying and actually are storing faces, but it doesn’t look like it. And it’s also perfectly valid to object to them storing even “male, 24”.
Yeah but this is the University of Waterloo we’re talking about here. This hit Canadian mainstream media CTV News so I know that. Also for an university specializing in Engineering and Mathematics there’s a shit ton of cameras around
Also, I’m not sure if this is the same in Canada as the US, but I’m pretty sure that in many cases, vandalism is considered a much lesser crime than unauthorized computer tampering/hacking
The students should get together and jack the machine away into their hacking club and do some reverse engineering, so that we get more information on how the data collection worked as opposed to just trusting the company’s statements. If a hacking group like the German Chaos Computer Club got behind this, they could release their findings while keeping the perpetrators anonymous. However, I’m pretty sure the machine is just a frontend to a server, which got shut down as soon as the students complained, with no GDPR-like checkout being available in the jurisdiction.
Vending machine in the Amazon warehouse in MN just had a camera taking pictures also. Fucking weird ass rich people/corps. A person was behind this decision and that face needs to be woken up with a punch everyday…
No only was a person behind the decision, a person was also behind the dissemination of the requirements, the implementation of the change, the design of the hardware, and all steps in between.
When you start tinkering with a machine learning model of any kind, you’re probably going to find some interesting edge cases the model can’t handle correctly. Maybe there’s a specific face that has an unexpected effect on the device. What if you could find a way to cheese a discount out of it or something?
Imagine a racist vending machine. The face recognition system think this customer is black with 81% confidence. Let’s increase the price of grape soda! Oh look, a 32 year old white woman (79% confidence). Better raise the price of diet coke!
In Japan they had some kind of facial recognition on vending machines selling cigarettes that would determine the age of the person in attempt to prevent kids from buying cigarettes. But it only worked for Japanese people.
Stupid racist vending machine wouldn’t sell me smokes!
Also I was in a diverse group of people and we were able to do some science. Fortunately we had a Japanese person in the group which allowed me to purchase the smokes. But yeah, it failed on everyone that wasn’t Japanese.
When you use a generated face with a mixture of white and black features, that’s when it gets interesting. Maybe you can even cause an integer overflow.
I don’t think they’re doing dynamic pricing on an individual basis, that would be too obvious. But checking the demographics of each location or individuals’ shopping habits, and potentially adjusting the prices or offerings? Definitely.
So, if you show it 100 faces from group A and 4 faces from group B, that could start gradually shifting the prices in a specific direction. If you keep going, you might be able to make it do something funny like charging 0.1 € for a Pepsi and 1000 € for a Coke or something like that. If the devs saw that coming, they might have set some limits so that the price can’s spiral totally out of control.
I am sure the profit margin is taken into account, so you won’t get an ultracheap Pepsi unless it expires soon. Similarly, I expect it to consider economic viability, so it won’t keep raising prices unless people are willing to pay them. Of course, you never know what the model actually does or what goals it follows (maximizing profit is a good guess, though), or how bad the coding is. The program might be very versatile and robust, or it may break when you show it a QR code - how can I know? Probably something in between.
I don’t get it either. What do vending machines need to be computerized for at all? What was wrong with the old kind that was around for decades where you put your money in, pushed a button, and stuff came out? I certainly can’t think of a reason for a vending machine to have a camera. That’s nuts.
It’s 2024, most people don’t carry cash, and the whole world runs on automation. These kinds of vending machines are completely over the top, but it’s actually a pretty bad idea to not use computers for this application. Just knowing when machines need to be refilled remotely saves more money than such an implementation would cost.
It’s not hard to know when machines need to be refilled. You just come regularly, take note of how much or little stock has been purchased, then adjust your refill amounts and times accordingly. This has to be done regardless of a handful of computerized machines because plenty of them still aren’t.
Accepting a credit card or tap-to-pay would probably require computerization, but the technology should be no more complex than any other, similar piece of hardware and the machine should even be able to work if the card network is down and just accept cash if that happens.
So sure, part of the machine should be computerized. The part that accepts money. The rest is unnecessary, probably raises the price of the machines unnecessarily and certainly never justifies a camera.
An IoT SIM costs a whole lot less than sending a technician to every machine to check stock. I’m not arguing in favor of facial recognition, I’ve already made that clear, but you are dead wrong if you don’t think automation at scale isn’t economical.
If you’re already putting a modem in the box for credit cards, why not collect some telemetry? Sensors are cheap and effective.
They have to go to every machine to restock regardless. All they have to do is note down on a little notepad or even an app on their phone what sells, what doesn’t and how quick.
I’m sorry, I just can’t go along with internet-connected public vending machines. If you want to connect everything in your house to the internet, fine. But a machine that sells candy bars does not need to be connected to the internet just because it’s marginally more efficient to do so than the way it had been done previously for decades. Because it results in this sort of shit. And unnecessary price-gouging through selling a university expensive machines with an unnecessary connection to the internet instead of something that worked perfectly well already and didn’t cost as much money.
They have to go to every machine to restock regardless.
But by managing stock over IoT one can minimize the amount of visits to only when machines need restocking, instead of also having to go to check stock
A $5 esp32 has more than enough computing power to run a vending machine, cameras and connectivity too. Would be cheaper than having to run all the analogue circuitry.
It’s not hard to know when machines need to be refilled. You just come regularly, take note of how much or little stock has been purchased, then adjust your refill amounts and times accordingly. This has to be done regardless of a handful of computerized machines because plenty of them still aren’t.
I worked in the arcade/vending business in the 1990s. That blind maintenance model was a crapshoot for the machine owners. We had to routinely send a crew (usually me and one other person) to drive to a location - near or far - with games, photo booths, vending stuff, etc. just in case the supplies in some machine or another ran out, something needed fixing, etc. Sometimes we’d arrive and learn we have hours of refilling and/or maintenance work to do on a machine, sometimes it had been a slow week or two and a crew had just spent their whole workday and a tank of gasoline to collect $50 from the cash box and go home again. Remote administration really changed the game for that whole business.
OS is a specific distribution, so like a Debian is the full fledged OS.
So just write your inventory inside the file, and bind the vending machine keys to it, and ignore 99% of the OS. The coin slot I would expect runs its own validations.
Why not? A full windows environment (though not really, because these things run what’s called the kiosk mode) can run on cheap SBCs and gives you a ton of hardware and software flexibility, and is also pretty convenient. It’s very commonly used for very good reasons.
Let’s stop kidding ourselves, the “good reason” is the cto’s yearly Microsoft financed holidays and/or too much legacy code to restart from scratch. But from a purely technical aspect, there’s no reason to touch windows
I can understand it from the perspective of the developers who need to implement all this crazy tracking/advertising/graphics functionality, but imo a vending machine should only do three things:
Let me see what is available (preferably using glass)
Accept payment
Give me what I paid for
Vending machines have done this for decades without requiring an operating system. Keep it simple!
You could make logistics simpler by giving these things networking capacity so you can remotely track their stock and cash levels.
If your software needs to run on multiple different device configurations, you can simplify development and deployment by letting the OS handle a lot of the low level stuff.
In other words, a simpler machine is not necessarily going to be simpler to operate for the company.
A low end Windows PC can be had very cheap these days. Why bother doing something proprietary, if you can just cobble together something from off the shelf parts?
This isn’t even remotely true. Everyone knows that if you’re trying to do a cheap embedded product, you use SBCs and Linux. Using Windows for these kinds of applications is almost always the result of a company having a contract with Microsoft that leads their development strategy towards Microsoft’s offerings rather than the best offerings.
Also, in what universe is a Linux platform more proprietary than Windows?
why do people think it’s okay to do this shit? if you’re coding facial recognition for a vending machine, that’s like 80 steps too far down the capitalism ladder
if you took this machine back to the 1920s and told people what it was doing, they’d shoot at it. and probably you
This is the result of capitalism - corporations (aka the rich selfish assholes running them) will always attempt to do horrible things to earn more money, so long as they can get away with it, and only perhaps pay relatively small fines. The people who did this face no jailtime, face no real consequences - this is what unregulated capitalism brings. Corporations should not have rights or protect the people who run them - the people who run them need to face prison and personal consequences. (edited for spelling and missing word)
In the article is a sound explanation: the machine is activated by detecting a human face looking at the display.
If this face recognition software only decides “face” or “not face” and does not store any data, I’m pretty sure this setup will be compatible with any data protection law.
OTOH they claim that these machines provide statistics about age and gender of customers. So they are obviously recognising more than just “face yes”. Still – if the data stored is just a statistics on age and gender and no personalised data, I’m pretty sure it still complies even with 1920s data protection habits.
I’m pretty sure that this would be GDPR conform, too, as long as the customer is informed, e.g. by including this info in the terms of service.
I don’t know about the US, but in Germany, by using a vending machine, you are implicitely and automatically consenting with the ToS of the vendor by your action.
I’ll play devil’s advocate. The machine recorded estimated age and gender. Assuming it tracked statistics and didn’t store images, what is the real harm? Future candy will have different designs after they found most users were 70yr old grandpas?
It is anonymized PII data collected without explicit consent, sure, but don’t blow it out of proportion. There is no big surveillance state plot here (yet), just an overzealous marketing team.
Not everybody who approaches the machine or walks past it is really consenting to their appearance being logged and analysed though - not to mention that “we don’t store data” is only true if the security is effective and no exploits manage to weaponise the camera now staring back at you as you try to make a purchase.
Ultimately vending machines are completely passive sales anyway, the collection of demographic data about who is buying from the machine are a little useless because it’s not like the machine can work on its closing techniques for coin based candy sales.
If you don’t have access to the source code then you don’t know what it’s doing. If there’s economic incentives to take my picture and tie my face to my name then I’m going to assume “trust us, it’s anonymous” means “we buy and sell your data” (at least).
If you’ll grant there are people in power who would want a surveillance state and businesses routinely sell data to governments then you don’t get to dismiss this out of hand. We have to draw the line somewhere, even if marketing people with a stalker mentally don’t see the line.
Add comment