Age Verification Cameras Easily Fooled


Nice try, Fujitaka. They manufacture cigarette vending machines in Japan, and were all set to roll out a new system of age verification cameras on their machines, which would scan the face of the buyer to look for sagging skin, wrinkles, age spots, and other signs of legal smoking age. The system is easily circumvented, however: people with a photo of an older person clipped from a magazine can fool the machine by simply showing the photo instead of their own faces.

Another aspect to Japan’s cigarette control is the Taspo card, which is an age verified ID issued to smokers of legal age or older (20 years old is the legal age in Japan). Taspo cards are required for over the counter purchases, and the majority of vending machines require them as well. Relatively few machines are outfitted with face recognition systems, but many more are set to ship in the coming months. Fujitaka claims they are working on a solution by improving the face recognition software, but we think it would be a lot easier to simply check the background of the image. Since the camera is static and always pointed in one direction, the portraits it captures should always have the same background. Someone please tell Fujitaka we just saved them a boatload of R&D money… until Guy Fawkes masks become more popular.

18 thoughts on “Age Verification Cameras Easily Fooled

  1. @Matthew

    It would only be a matter of time until they start circulating a phone app to circumvent it. Just find an old guy to stand in front of a bluescreen, record him making a series of goofy faces and assign the resulting animations to keys on a smartphone. then just snap a picture of the area in front of the machine to fool juan’s background recongnition and collect your smokes.

    It could be made more elaborate from that point, like using a david-style cam and laser setup to map a face to 3D and check for agedness, but then you’d just have teens buying latex masks. I suppose you could circumvent even that by updating the smartphone app with red lines, but then they’d just start randomly generating and watching their scanlines. That’d probably hog the processor enough that they wouldn’t look for serial smiles and frowns, so latex masks would just find their way into the hands of schoolchildren..

  2. I like how in America we just get rid of the vending machines to prevent underage smoking and rely on a person, but in Japan they simply try to turn them into super intelligent robots.

  3. I’ve seen infrared cameras snap pictures of the veins in a person’s hand for identify verification. The infrared cameras could be used to verify that the image is real by looking for veins in the person’s face. I could imagine this being quite tricky to circumvent.

  4. Instead of guy fawkes masks, kids might start wearing Nixon masks. What happens to people who don’t have wrinkled skin, will the machines not sell cigarettes to them?

  5. 7:
    Don’t know how common it is, but some veins in my hand can be “pushed around” as if they aren’t attached inside, just sort of loose there. I guess thats going to confuse the cameras quite a bit.

  6. This is all good and well, but you have to understand culture in Japan is much more honest and trustworthy than in the US. In my opinion the likeliness of a Japanese teen going out of their way to cheat over a vending machine is not that much. All that, is part of the reason you can buy alcohol and almost anything out of vending machines in the streets of Japan. Also, the newer machines might be different but to back up my point, a lot of times you would just leave the money in the machine and open it up and take your beverage etc. In contrast, with their westernizing culture, this is probably more and more necessary.

  7. ehrm…why so complicated?? here in austria, you need your bank card to buy cigarettes at the vending machines..they have your date of birth stored on them, so finding out if youre the right age is a matter of adding integers… do the japanese have too many bored engineers?? if so then why aren’t they working on flying cars instead?? seriously..facial feature recognition has to be THE most inconvenient way to confirm the buyers age…

  8. Living here, and seeing the machines (i dont smoke, but I do drink) I think facial recognition is over the top. Using the card is good enough (the only way to get one is at a cig store or government office, where a person actually verifies your age.
    To combat theft and cloaning/faking cards, why not use thumbprint scan or vein scan instead of facial? Yeah, again, both of them could be hacked (as can *anything*) but a combination of both card and thumbprint would make it seriously inconvienient.

    as to the lengths kids will go to here in such an *honest* culture (remember, america is invading japan), I wouldnt put it past them.. they are technically proficient, cleaver, and willing to work a little for illegal stuff.
    Case in point: its well known that kids shoplift books from new bookstores, then sell them to used bookstores for cash here. A little effort and a desire and kids will do anything.. no matter the country.

  9. Err, what about the other direction–people that just look young? I know lots of folks who absolutely look 10 years younger than they are. There is no way machine vision would/could be better at interpreting human facial features than an actual human…

  10. just because the camera is pointed in one direction does not make the background static. surroundings change other people can be behind in the background. nothing stays the same.. horrible suggestion hackaday you didnt save them any money.

    I haven’t been on hackaday in a while and i have come back to find it is commercialized and idiotic, with very few ingenius “hacks”

    Hackaday has become a news site for stuff that doesn’t even get on digg these days. and i dont mean the good stuff that used to.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.