In Soviet Russia, Doorbell Rings You

We can imagine that the origin of the doorbell is truly ancient. if you lived in a cave, you probably had a stick and a rock nearby for people to get your attention without invading your cave. In 1817 a Scot named William Murdoch had a bell in the house that visitors rang via a compressed air system, but the electric doorbell had to wait until 1831. Since then, little has changed with the basic idea. [Erientes] — who lives in the Netherlands, not Russia — wanted a smarter doorbell. In particular, he’s read about older people being victimized by people who ring the doorbell for entry. So [Erientes] used a Raspberry Pi to make a doorbell that supports facial recognition.

The exercise is really more of an operations challenge than a technical one thanks to a high-quality Python library for face recognition powered by DLib. However, we did like the user interface aimed at non-technical users. The metaphor is a traffic light in which a red light means do not allow entry. The lights are buttons, so you can use them to whitelist or blacklist a particular person.

Continue reading “In Soviet Russia, Doorbell Rings You”

Hackaday Links: November 10, 2019

In the leafy suburbs of northern Virginia, a place ruled by homeowner’s associations with tremendous power to dictate everything from the color of one’s front door to the length of grass in the lawn, something as heinous as garage doors suddenly failing to open on command is sure to cause a kerfuffle. We’ve seen this sort of thing before, where errant RF emissions cause unintentional interference, and such stories aren’t terribly interesting because the FCC usually steps in and clears things up. But this story is a little spicier given the source of the interference: Warrenton Training Center, a classified US government communications station located adjacent to the afflicted neighborhood. WTC is known to be a CIA signals intelligence station, home to spooks doing spooky stuff, including running high-power numbers stations. The interference isn’t caused by anything as cloak-and-dagger as that, though; rather, it comes from new land-mobile radios that the Department of Defense is deploying. The new radios use the 380-400 MHz band, which is allocated to the Federal Government and unlicensed Part 15 devices, like garage door remotes. But Part 15 rules, which are clearly printed on every device covered by them, state that the devices have to accept unwanted interference, even when it causes a malfunction. So the HOA members who are up in arms and demanding that the government buy them new garage door openers are likely to be disappointed.

Speaking of spooks, if you’re tired of the prying electronic eyes of facial recognition cameras spoiling your illusion of anonymity, have we got a solution for you. The Opt-Out Cap is the low-tech way to instantly change your face for a better one, or at least one that’s tied to someone else. In a move which is sure not to arouse suspicion in public, doffing the baseball cap deploys a three-piece curtain of semi-opaque fabric, upon which is printed the visage of someone who totally doesn’t look creepy or sketchy in any way. Complete instructions are provided if you want to make one before your next trip to the ATM.

It’s always a great day when a new Ken Shirriff post pops up in our feed, and his latest post is no exception. In it, Ken goes into great detail about the history of the 80×24 (or 25) line standard for displays. While that may sound a bit dry, it’s anything but. After dispelling some of the myths and questionable theories of the format’s origin – sorry, it’s not just because punch cards had 80 columns – he discusses the transition from teletypes to CRTs, focusing on the very cool IBM 2260 Display Station. This interesting beast used an acoustic delay line made of 50′ (15 m) of nickel wire. It stored data as a train of sound pulses traveling down the wire, which worked well and was far cheaper than core memory, even if it was susceptible to vibrations from people walking by it and needed a two-hour warm-up period before use. It’s a fascinating bit of retrocomputing history.

A quick mention of a contest we just heard about that might be right up your alley: the Tech To Protect coding challenge is going on now. Focused on applications for public safety and first responders, the online coding challenge addresses ten different areas, such as mapping LTE network coverage to aid first responders or using augmented reality while extricating car crash victims. It’s interesting stuff, but if you’re interested you’ll have to hurry – the deadline is November 15.

And finally, Supercon starts this week! It’s going to be a blast, and the excitement to hack all the badges and see all the talks is building rapidly. We know not everyone can go, and if you’re going to miss it, we feel for you. Don’t forget that you can still participate vicariously through our livestream. We’ll also be tweet-storming and running a continuous chat on Hackaday.io to keep everyone looped in.

Be Anyone Or Anything With Facial Projection Mask

In the market for a low-poly change to your look? Hate the idea of showing up for a costume party only to find out someone is wearing the same mask as you? Then this face changing front-projection mask may be just the thing for you.

To be honest, we’re not sure just how much [Sean Hodgins]’ latest project has to do with cosplay. He seems to be making a subtle commentary about dealing with life in the surveillance state, even though this is probably not a strategy for thwarting facial-recognition cameras. [Ed Note: Or maybe it’s just Halloween?]

The build consists of a Raspberry Pi and a pico projector of the kind we’ve seen before. These are mated together via a custom PCB and live inside a small enclosure that’s attached to the end of a longish boom. The boom attaches to the chin of 3D-printed mask, which in turn is connected to the suspension system of a welding helmet. Powered by a battery pack and controlled by a smartphone app, the projector throws whatever you want onto the mask – videos, effects, even images of other people. Even with some Photoshop tweaks to account for keystone distortion from the low angle of projection, there’s enough distortion that the effect is more artistic than masquerade. But honestly, having your face suddenly burst into flames is pretty cool. We just wonder what visibility is like for the wearer with a bright LED blasting into your eyes.

As a bonus, [Sean] has worked this build into a virtual treasure hunt. Check out 13thkey.com and see what you can make from the minimal clues there.

Continue reading “Be Anyone Or Anything With Facial Projection Mask”

Hackaday Links: October 13, 2019

Trouble in the Golden State this week, as parts of California were subjected to planned blackouts. Intended to prevent a repeat of last year’s deadly wildfires, which were tied in part to defective electrical distribution equipment, the blackouts could plunge millions in the counties surrounding Sacramento into the dark for days. Schools have canceled classes, the few stores that are open are taking cash only, and hospitals are running on generators. It seems a drastic move for PG&E, the utility that promptly went into bankruptcy after being blamed for last year’s fires, but it has the support of the governor, so the plan is likely to continue as long as the winds do. One group is not likely to complain, though;  California amateur radio operators must be enjoying a greatly decreased noise floor in the blackout areas, thanks to the loss of millions of switch-mode power supplies and their RF noise.

Good news, bad news for Fusion 360 users. Autodesk, the company behind the popular and remarkably capable CAD/CAM/CAE package, has announced changes to its licensing scheme, which went into effect this week. Users no longer have to pay for the “Ultimate” license tier to get goodies like 5-axis machining and generative design tools, as all capabilities are now included in the single paid version of Fusion 360. That’s good because plenty of users were unwilling to bump their $310 annual “Standard” license fee up to $1535 to get those features, but it’s bad because now the annual rate goes to $495. In a nice nod to the current userbase, those currently on the Standard license, as well as early adopters, will get to keep the $310 annual rate as long as they renew, and The $495 pricing tier went into effect in November of 2018, while anyone still on the $310 annual price was grandfathered in (and will remain to be). At that time there was still a $1535 tier called Ultimate, whose price will now be going away but the features remain in the $495 tier which is now the only pricing option for Fusion 360. Ultimate users will see a $1040 price drop. As for the current base of freeloaders like yours truly, fear not: Fusion 360 is still free for personal, non-commercial use. No generative design or tech support for us, though. (Editor’s Note: This paragraph was updated on 10/14/2019 to clarify the tier changes after Autodesk reached out to Hackaday via email.)

You might have had a bad day at the bench, but was it as bad as Román’s? He tipped us off to his nightmare of running into defective Wemos D1 boards – a lot of them. The 50 boards were to satisfy an order of data loggers for a customer, but all the boards seemed caught in an endless reboot loop when plugged into a USB port for programming. He changed PCs, changed cables, but nothing worked to stop the cycle except for one thing: touching the metal case of the module. His write up goes through all the dead-ends he went down to fix the problem, which ended up being a capacitor between the antenna and ground. Was it supposed to be there? Who knows, because once that cap was removed, the boards worked fine. Hats off to Román for troubleshooting this and sharing the results with us.

Ever since giving up their “Don’t be evil” schtick, Google seems to have really embraced the alternative. Now they’re in trouble for targeting the homeless in their quest for facial recognition data. The “volunteer research studies” consisted of playing what Google contractors were trained to describe as a “mini-game” on a modified smartphone, which captured video of the player’s face. Participants were compensated with $5 Starbucks gift cards but were not told that video was being captured, and if asked, contractors were allegedly trained to lie about that. Contractors were also allegedly trained to seek out people with dark skin, ostensibly to improve facial recognition algorithms that notoriously have a hard time with darker complexions. To be fair, the homeless were not exclusively targeted; college students were also given gift cards in exchange for their facial data.

For most of us, 3D-printing is a hobby, or at least in service of other hobbies. Few of us make a living at it, but professionals who do are often a great source of tips and tricks. One such pro is industrial designer Eric Strebel, who recently posted a video of his 3D-printing pro-tips. A lot of it is concerned with post-processing prints, like using a cake decorator’s spatula to pry prints off the bed, or the use of card scrapers and dental chisels to clean up prints. But the money tip from this video is the rolling cart he made for his Ultimaker. With the printer on top and storage below, it’s a great way to free up some bench space.

And finally, have you ever wondered how we hackers will rebuild society once the apocalypse hits and mutant zombie biker gangs roam the Earth? If so, then you need to check out Collapse OS, the operating system for an uncertain future. Designed to be as self-contained as possible, Collapse OS is intended to run on “field expedient” computers, cobbled together from whatever e-waste can be scrounged, as long as it includes a Z80 microprocessor. The OS has been tested on an RC2014 and a Sega Master System so far, but keep an eye out for TRS-80s, Kaypros, and the odd TI-84 graphing calculator as you pick through the remains of civilization.

The Tens Of Millions Of Faces Training Facial Recognition; You’ll Soon Be Able To Search For Yourself

In a stiflingly hot lecture tent at CCCamp on Friday, Adam Harvey took to the stage to discuss the huge data sets being used by groups around the world to train facial recognition software. These faces come from a variety of sources and soon Adam and his research collaborator Jules LaPlace will release a tool that makes these dataset searchable allowing you to figure out if your face is among the horde.

Facial recognition is the new hotness, recently bubbling up to the consciousness of the general public. In fact, when boarding a flight from Detroit to Amsterdam earlier this week I was required to board the plane not by showing a passport or boarding pass, but by pausing in front of a facial recognition camera which subsequently printed out a piece of paper with my name and seat number on it (although it appears I could have opted out, that was not disclosed by Delta Airlines staff the time). Anecdotally this gives passengers the feeling that facial recognition is robust and mature, but Adam mentions that this not the case and that removed from highly controlled environments the accuracy of recognition is closer to an abysmal 2%.

Images are only effective in these datasets when the interocular distance (the distance between the pupils of your eyes) is a minimum of 40 pixels. But over the years this minimum resolution has been moving higher and higher, with the current standard trending toward 300 pixels. The increase is not surprising as it follows a similar curve to the resolution available from digital cameras. The number of faces available in data sets has also increased along a similar curve over the years.

Adam’s talk recounted the availability of face and person recognition datasets and it was a wild ride. Of note are data sets by the names of Brainwash Cafe, Duke MTMC (multi-tracking-multi-camera),  Microsoft Celeb, Oxford Town Centre, and the Unconstrained College Students data set. Faces in these databases were harvested without consent and that has led to four of them being removed, but of course, they’re still available as what is once on the Internet may never die.

The Microsoft Celeb set is particularly egregious as it used the Bing search engine to harvest faces (oh my!) and has associated names with them. Lest you think you’re not a celeb and therefore safe, in this case celeb means anyone who has an internet presence. That’s about 10 million faces. Adam used two examples of past CCCamp talk videos that were used as a source for adding the speakers’ faces to the dataset. It’s possible that this is in violation of GDPR so we can expect to see legal action in the not too distant future.

Your face might be in a dataset, so what? In their research, Adam and Jules tracked geographic locations and other data to establish who has downloaded and is likely using these sets to train facial recognition AI. It’s no surprise that the National University of Defense Technology in China is among the downloaders. In the case of US intelligence organizations, it’s easier much easier to know they’re using some of the sets because they funded some of the research through organizations like the IARPA. These sets are being used to train up military-grade face recognition.

What are we to do about this? Unfortunately what’s done is done, but we do have options moving forward. Be careful of how you license images you upload — substantial data was harvested through loopholes in licenses on platforms like Flickr, or by agreeing to use through EULAs on platforms like Facebook. Adam’s advice is to stop populating the internet with faces, which is why I’ve covered his with the Jolly Wrencher above. Alternatively, you can limit image resolution so interocular distance is below the forty-pixel threshold. He also advocates for changes to Creative Commons that let you choose to grant or withhold use of your images in train sets like these.

Adam’s talk, MegaPixels: Face Recognition Training Datasets, will be available to view online by the time this article is published.

AI On Raspberry Pi With The Intel Neural Compute Stick

I’ve always been fascinated by AI and machine learning. Google TensorFlow offers tutorials and has been on my ‘to-learn’ list since it was first released, although I always seem to neglect it in favor of the shiniest new embedded platform.

Last July, I took note when Intel released the Neural Compute Stick. It looked like an oversized USB stick, and acted as an accelerator for local AI applications, especially machine vision. I thought it was a pretty neat idea: it allowed me to test out AI applications on embedded systems at a power cost of about 1W. It requires pre-trained models, but there are enough of them available now to do some interesting things.

You can add a few of them in a hub for parallel tasks. Image credit Intel Corporation.

I wasn’t convinced I would get great performance out of it, and forgot about it until last November when they released an improved version. Unambiguously named the ‘Neural Compute Stick 2’ (NCS2), it was reasonably priced and promised a 6-8x performance increase over the last model, so I decided to give it a try to see how well it worked.

 

I took a few days off work around Christmas to set up Intel’s OpenVino Toolkit on my laptop. The installation script provided by Intel wasn’t particularly user-friendly, but it worked well enough and included several example applications I could use to test performance. I found that face detection was possible with my webcam in near real-time (something like 19 FPS), and pose detection at about 3 FPS. So in accordance with the holiday spirit, it knows when I am sleeping, and knows when I’m awake.

That was promising, but the NCS2 was marketed as allowing AI processing on edge computing devices. I set about installing it on the Raspberry Pi 3 Model B+ and compiling the application samples to see if it worked better than previous methods. This turned out to be more difficult than I expected, and the main goal of this article is to share the process I followed and save some of you a little frustration.

Continue reading “AI On Raspberry Pi With The Intel Neural Compute Stick”

Hackaday Podcast Ep1 – Seriously, We Know What We’re Doing

First podcast of the new year! Editors Elliot Williams and Mike Szczys look back on the most interesting hacks and can’t-miss articles from the past week (or so). Highlights include abusing IPv6 addresses, underclocking WiFi, taking Wii out of the livingroom, scratch built microphones, computer prophecy coming true, and the end of an automotive era. Full show notes below.

This week, Hackaday Contributor Bob Baddeley came on the show to discuss developments in facial recognition technology and its use in the wild.

Direct Download (45.1 MB MP3)

Places to follow Hackaday podcasts:

Continue reading “Hackaday Podcast Ep1 – Seriously, We Know What We’re Doing”