It appears there's a terribly contagious case of switcheroo fever in the comment section here.
The objective of the hoodie is to prevent face recognition.
Those who have come down with switcheroo fever are saying "aha! This actually makes you more recognizable, not less!"
But this is not a trick question and this is not the type of thing where the switcheroo is meaningfully responsive.
The objective is to prevent a camera from recognizing your face, and it does that. If you have tested positive for switcheroo fever, you should start your comment by stating out loud that you understand this is the objective and you understand that it does indeed achieve the objective before proceeding elaborate on your own personal variation of the switcheroo. And perhaps, if you dare, proceed to challenge mode, where you acknowledge that this might be a trade-off that was consciously chosen, and does not come as a surprise to the people who designed the product.
Why should commenters be constrained to only comment on the stated objective of the hoodie, and not all of its pros and cons?
> If you have tested positive for switcheroo fever, you should start your comment by stating out loud that you understand this is the objective and you understand that it does indeed achieve the objective before proceeding elaborate on your own personal variation of the switcheroo.
I don't think this is necessary. If someone doesn't make any mention of whether or not it succeeds at preventing facial recognition, then they are simply not offering an opinion on the matter. Not mentioning it, or bringing up a different concern, should not be taken as disagreement. In fact, I would interpret silence on the matter as a sort of implicit agreement.
> And perhaps, if you dare, proceed to challenge mode, where you acknowledge that this might be a trade-off that was consciously chosen, and does not come as a surprise to the people who designed the product.
So commenters here are obligated to acknowledge a bunch of considerations about the designer's objective and state of mind when creating the hoodie, but the designer is not obligated to point out these trade-offs in their product listing? Isn't this a double standard?
I think most of the people talking about how it makes you more recognizable don't understand how security cameras actually work. Most of the time, nobody is actually watching the camera. It's there to have a recording of any incident after it happens.
The IR hats that were going around the internet 10 years ago, and this modern version (in a hoodie) are about defeating that security model. The wearer's face becomes completely unrecognizable in recordings, and the wearer has plausible deniability about being at the location if questioned after the fact.
When security personnel are actively watching the cameras, like at a high-security facility, this makes you stand out. But if you're at a convenience store, it completely defeats their camera recordings.
> But if you're at a convenience store, it completely defeats their camera recordings.
Not if the camera has an IR-cut filter. And don't security camera feeds frequently get displayed to employees behind the counter (contrary to your assertion that nobody is watching)?
A quick search for "IR cut security camera" on Amazon returns hundreds of results. Some cost as little as $20 [1], so not uncommon at all from what I can tell.
> Even for those that do, IR filters are rarely perfect, so a bright enough IR light would still do the trick.
Do you have a source for this? I've never tested it, so I can't say, but I would want to want to see some proof first.
> A quick search for "IR cut security camera" on Amazon returns hundreds of results. Some cost as little as $20 [1], so not uncommon at all from what I can tell.
That's the very opposite. That's a camera that uses IR illumination to record at night. That's what all those LEDs around the lens are for. And due to this it lacks any IR filtering. This particular camera is more IR sensitive than most cameras like DSLRs or cell phones.
> Do you have a source for this? I've never tested it, so I can't say, but I would want to want to see some proof first.
You can test it by pointing a remote at a camera. Normal cameras like cell phones and DSLRs have IR filters. They still can see the blinking LED on an IR remote, because the filter isn't perfect.
> That's the very opposite. That's a camera that uses IR illumination to record at night. That's what all those LEDs around the lens are for. And due to this it lacks any IR filtering. This particular camera is more IR sensitive than most cameras like DSLRs or cell phones.
No, the camera also has IR cut - it lists the feature right in the product title, and the customer-provided samples in the reviews clearly have an IR cut filter applied (you can tell by the "natural" colors in the daytime photos). Is has two modes of operation, daytime mode and nighttime mode. During daytime mode it employs an IR cut filter, and during nighttime mode it removes the filter and adds additional IR lighting for "night vision".
> They still can see the blinking LED on an IR remote, because the filter isn't perfect.
Being able to see a faint blinking LED from a remote is not sufficient to prove that the effect is strong enough to completely obscure a person's face when an IR cut filter is applied, in my opinion.
> During daytime mode it employs an IR cut filter, and during nighttime mode it removes the filter and adds additional IR lighting for "night vision".
Ah, I see
> Being able to see a faint blinking LED from a remote is not sufficient to prove that the effect is strong enough to completely obscure a person's face when an IR cut filter is applied, in my opinion.
It's just a matter of intensity. An IR filter is going to block some percentage. If it blocks 90%, you just need a 10X stronger light. At some point this gets impractical of course, but you can buy some really powerful LEDs. Most security cameras are cheap and the IR filter is there to improve the image in the daytime, not to counteract intentional attacks.
If I'm working at a convenience store, and I notice that someone walked in wearing a camera jamming device, I am going to be pretty suspicious. Almost as suspicious as if they came in wearing a ski mask. Not sure what I'd do about it, but I would definitely give a fuck.
Having worked retail I can tell you the video feed is interesting for all of about 10 minutes. After that, there is a job to do. The only time it gets looked at is if a hot girl or some sketchy AF walks into the store and down tall isles. In practical terms, no one is watching the camera feed after their first day on the job.
Almost every camera has an IR filter and can still see high-intensity IR light. Point your phone camera at a TV remote and you can see this. That camera has a strong IR filter, but the little LED in the remote overwhelms it still.
Maybe the name "camera-shy" is not the best description of something that actively goes out and targets cameras specifically, floods them with signals of its own, and therefore stands out.
Camera-jamming or IR-glare or something on those lines may convey the function better without setting wrong expectations.
> The objective is to prevent a camera from recognizing your face
Not just "a camera", but one that's in night-vision mode. The camera may not be using IR, or it may switch to normal mode at any time. You spent too much time telling people off for pointing out this hoody makes your head glow and may draw attention as a consequence.
Not necessarily for either point, much as I want to avoid being pedantic.
My LG CX's remote uses IR for turning on, but beyond that it uses Bluetooth or some similar protocol.
Even if a remote is IR, rear cameras are often shielded against it. Front cameras, usually not. This is much easier to notice on IR-powered PC mice of course. Not sure if modern ones use IR as well - haven't checked.
I think there's a misconception here: most security setups aren't like in the movies where there's a guy sitting in a room with 20 screens, watching all the security camera feeds intently for something out of the ordinary. Yes, in high-security facilities, that's probably the case, but for the vast majority of security cameras, no one is monitoring them in real-time. They're just there so that when something does happen, someone can go look at the recorded footage to see if there are any clues or evidence.
The "camera-shy hoodie" defeats this latter type of security camera setup. Perhaps the article should explicitly say that this hoodie will be counter-productive in a high-security facility, as you'll be immediately noticeable to someone watching those cameras in real-time. But most cameras in the real world will not be monitored in real-time.
Would depend why one doesn't want to be seen by cameras I suppose. I imagine in many activities where one would want to avoid being recorded, one would also want to avoid wearing the most visible and noticeable garment possible.
I see a bunch of people wearing these, and when driving at night, they become really really visible, in a good way. They're less bright than lights from oncoming traffic, so not an issue of being 'too bright', but you can still see the pedestrian from far away, realize it's a human (a lot easier than with reflective strips or dots), and drive accordingly.
10/10, would recommend to pedestrians walking on the steet at night!
When I used to ride a motorcycle, I remember lots of riders used to advise having your full lights on when driving during the day, so you were more noticeable to motorists.
However, I noticed a driver holding up their hand to shield their eyes while I was riding during the day.
Needless to say I didn't do that any longer after that.
Drivers may develop target fixation and inadvertently steer towards something they want to avoid. It's thought to be one of the reasons that so many cars manage to nail light poles and trees dead-center.
When they've already lost most control and are panicking, yes. But it's completely normal for cyclists to wear reflective clothing to make themselves visible. Even cars have retroflectors on their taillights for this purpose. No road user is trying to hide in the dark in the hopes that drivers won't get fixated on them!
The reflections from these jackets are not dazzling so much as ... broad spectrum? Basically they just look like a bright white sheet. Very hard to miss, hopefully.
The idea of a Tesla self-driving system getting target fixation and driving towards the thing it should avoid is hilarious to me. And it also makes me wonder how likely that is to actually happen. Is the model weight of "shiny, bright reflective thing in IR spectrum" for "probably a road reflector, avoid" higher than "probably some water in the road, so that must be where the road is"? (Obviously totally made up examples).
Uber ran over a woman [1] walking with a bicycle because she "didn't exist" as far as their software was concerned. This is a real danger when using the public as beta testers for robots. The Uber vehicle has more sophisticated sensors than a Tesla which is winging it with just cameras.
I was curious about the result of the investigation since the article you linked said it was not yet known whether the self-driving features of the Tesla were active during the accident.
The outcome of the investigation was that the Autopilot was not engaged and the cause of this particular crash was human error:
"Per the investigation, the car data showed that the driver hit the gas pedal rather than the brake. The Model S then entered the rest stop, going 60 mph, hitting a curb and then the trailer."
Someone can wish no ill to others, yet still find irony or absurdity in things that already happened. It's something that both victims and witnesses often do as a way of processing life's darker realities. Please don't tell other people how they should experience tragedy or speak on behalf of victims. You're no better situated to do that the person you responded to.
>Is there any downside to these? Such as distracting cars or blinding other cyclists?
One speculative possible downside: others have answered the human aspect, but in the context of security/privacy OP is being perhaps a little blase with "plausible deniability", or at least it may vary by region? I can see why such a jacket could be useful and why a cyclist or jogger or the like would wear one, same as other visibility enhancing gear. But at the same time I've never ever seen such a thing anywhere in our state. And on camera it apparently looks like a neon sign. Sure, that obscures your face, but at the same time it will instantly draw the attention of any human security watching or potentially of AI if it's been tasked to look for such outliers. If it caught on and a quarter or half or whatever people were wearing such things at night then that wouldn't matter again, but anything that makes one "unique" in a crowd, regardless of what it is, represents signal. If someone is the only one (or one of only a couple) wearing something like that then even without their face they could still be tracked uniquely just by that. And with advances in later generation biometrics like gait analysis wearing such a thing might not only fail to stop individualized identification, but also leak information in terms of "this person is trying to hide their identity" which might correlate to groups in ways hard to predict.
Of course again, that could be taken as an argument to try to make all this sort of thing more of a trend, but there is a first-mover calculation to make maybe depending on what someone is trying to accomplish. If one just wants to help hide others or mess around with camera tracking for the fun of it though that'd be no problem. It's the same kind of dilemma as all sorts of other "blend with the crowd" stuff, like Tor say. If everyone ran an onion router (ignoring current gen scaling issues) then there'd be zero signal from doing so, the protection offered would be strongest, and also the vast majority of traffic would be completely boring and legitimate because so would almost all the users. It'd be much harder to crack down on as well. If only a few people run it and specifically do so to conduct covert activity, well the mere act of doing so then potentially leaks a bit. And both sides are subject to reinforcing effects, if "everyone does it" and "it's normal" and "almost entirely legit" then that makes it easy for anyone new to join, constantly reinforcing the network. Conversely if it's something strange with whiffs of criminality, then many won't which in turn reinforces it being strange and those using it maybe having something suspicious going on. Bootstrapping that can be a tough nut.
Anyway just sort of thinking out loud so to speak. For cameras I'm leaning towards ultimately some sort of legal response being needed, and that individuals trying to evade it is probably a losing battle.
Indeed, plausible deniability tends to be something they helps you blend in with the crowd, not simply having a legitimate excuse when it shines a bright light on you. It would likely be a net benefit to become common among pedestrians and cyclists regardless.
Catching criminals is a rarely a sufficient excuse for degradation of privacy for everyone.
I helped put up trail markers on backcountry ski trails, we used orange 3M ones with the micro cube corners that reflected an insane amount of light, life savers when in the middle of a storm. I always wondered what a jacket out of that material would be like, thanks for the link!
Wow, for a minute I thought someone did a low effort superimposition of a logo of the jacket or something into the image. Took me a moment to work out that scene.
Protest organizers should hand out these hoodies, or someone should sell them to protestors through some kind of pop-up shop ambassador program where an affiliate at each protest sells the apparel on consignment.
FYI check your local laws, this is almost certainly criminal in at least some places (e.g. in Germany it's illegal to "wear things which may prevent or interfere with identification of protestors" and to "carry items meant to interfere with identification", this clearly meets those criteria).
German police usually has a bunch of people (and cars with pole-mounted cameras) running around with cameras filming everyone at regular protests; you'd think nothing can happen without being taped. Except of course it often goes <FOOTAGE MISSING> when it gets interesting, seems to be a technical issue. Just like bodycams suddenly failing to work. Oh well
The primary qualifier here is this isn't American police. Use of force, excessive use of force, and lethal use of force are much, much rarer here. Police kills about a dozen people per year. Not over a thousand. Though worryingly this number has risen a lot in the last couple years, it used to be more like half a dozen ten years ago.
Regardless, when it happens, it's still a police force, so you still have esprit de corps, "blue wall of silence" and "DA doesn't feel like prosecuting people they work with day to day" effects, and evidence is quite frequently not gathered or "gets lost". So for example if there is excessive use of force at a protest it wouldn't be unusual for those recordings to be misplaced. Bodycams aren't universal yet, but they're always somehow turned off during excessive use of force. The latest example would be a 16yo refugee who was shot dead with an MP5 (i.e. not standard-issue-24/7-belt-carry for police, something you have to get from the car), all officers had bodycams, no recordings. Five officers were charged with various things (up to 2nd degree homicide) because a citizen made an audio recording of the police action (which is a felony btw.), revealing contradictory and false statements.
My googling has been failing to find info about the bodycams etc in this case. Do you have a link?
Additionally, do you have any figures, even very approximate, for what proportion of the time police bodycams are worn and on and what proportion of incidents have said camera turned off? This helps see if there's a real mismatch/incidence of corruption here.
With that being said, it is not illegal to attempt / to succeed at escaping a imprisonment as the will to be free is an unalienable human right under German law.
In Germany? “Some people in the crowd started to mask their faces“ is THE excuse for authorities to stop protests. You cannot seriously argue this doesn’t happen at e.g. 1st of May protests.
[0] I mean that in a legal sense, that's how the police works. Whether stopping the protest is morally the right thing to do or not is a different issue that I don't want to address here.
If you think German police(*) of all police forces goes easy on lefties you are completely delusional.
(*) The generalization is okay in this context, because behavior is pretty uniform and larger protests will have riot police pulled from other states anyway.
It is honestly just a terrible idea - you are essentially placing a target on yourself, no matter how you use it. Simply scanning a crowd of people with a camera, you are identifying people protesting from other civilians, in real-time and no cost. If you watched a crowd from above, over time you could find all sorts of patterns. In some countries, that could mean your whole family disappears.
The only groups I see benefitting from this are the agents of oppressive regimes, and the meth heads breaking into your storage shed at night and stealing your bike.
Depending on where you live, wearing a mask (as in: face covering, not the COVID masks) at a public protest or demonstration can be a punishable offense forbidden by law… and I'm not talking about autocratic states but many civilized and relatively liberal Western countries, cf https://en.wikipedia.org/wiki/Anti-mask_law
> For example, The Georgia Supreme Court found the law constitutional on the grounds that the wearing of the mask was an act of intimidation and a threat of violence
Women wear masks ever day with their make up, and also documented with filters on social media. Bet the courts didnt think of masks in that context did they?
I didnt get taught law, its not something that is a compulsory subject to teach and the adult population certainly doesnt get taught about the upgrades or changes to law, so I cant have a democratic debate over the laws that exist.
Ergo we are living in a legal dictatorship that put Mussolini to shame, but as most people dont know who he is, thats a point thats wasted on people.
Why is the state hell bent on removing everyone's privacy as well?
These anti-mask laws are designed to harm the public, especially when considering the riot police will be balaclava'ed up and in full riot gear.
If the scientific community especially at Uni, werent such pussy's there would be psychological paper's that exist demonstrating the provocative nature of the riot gear uniform, using military techniques, because the police and military share many training elements.
Representative democracy is a trick to cede autonomy over one's actions, so if you needed a way to control a population en-masse, this would be a disarmingly charming way to achieve this objective, the sort of thing a self attracted sociopath would come up with, and something that continues to attract sociopaths with narcissistic attributes around the world to this day!
I don't see where you would get that from my comment, which didn't name any country. The point was that the wiki article I pointed to does list quite a number of countries, of which several do fall into the group of “liberal Western countries”. Nobody said that _all_ countries listed there do.
The documentation lists its limitations, which include daylight:
> The Camera Shy Hoodie is not an end-all-be-all for hiding your identity. It’s good for one thing, blowing out the view of night vision (IR) cameras in low light environments. It’s not effective in sunlight, most indoor lighting, or against conventional cameras. In fact, you will draw attention to yourself if you wear this in a context in which the security
cameras are actively monitored. In the view of an IR camera, it’ll look as if you’re flashing a light directly at the lens. In addition, the LEDs need to be a minimum distance from the camera to be effective, as the cone of light from them needs to be wide enough to overlap with the cameras view of the wearers head. In practice, this is about 12ft (~3m).
The idea is designed to exploit limitations with the camera's auto-exposure, and I presume a requirement to do that is to strobe a light that outshines the ambient light by n times. That works at night because there's so little ambient light, during the day you'd have to outshine the sun (and the sun is really bright).
Be realistic, 'everyone should wear this to provide cover to people in stealth mode' is not a sensible security plan outside of an extremely limited number of contexts. I hate security camera theater and panopticons most than most people, and I have exactly zero expectation of this becoming A Thing.
Indeed, this is only really useful if the cameras aren't actively being monitored. If you're the odd one out wearing this at a protest, you might be specifically targeted.
Thats assuming they can figure out which one is wearing the hoodie. With enough evidence I can see them going for a guy with that, but they would have to not only make the decision to specifically target you, but to figure out which one of the hoodies moving around is causing the issue
you don’t have to commit a crime to wear this. you can wear it as a statement, in which case attracting that kind of attention so that you can say — ideally to someone with authority — “i don’t think it’s right to surveil the public like this” is arguably some measure of success.
Given that the cops have never needed much of a reason to illegally beat people up, I admire anyone who cares about making that statement so much that they're willing to risk it.
I think in the scenario this is trying to defeat, they already know who most of the other people are because their facial recognition software worked & they're already looking at their socials, purchase history, etc. All the 3 letter agencies are using this info, plus whatever else they have to construct some kind of case against them.
So, they become the low hanging fruit. The more resources we give our TLA's, the less this matters, of course :)
That's why you put a pebble in your shoe. Then after enough years you take it out like Piccolo dropping his weighted clothes and, like Harrison Bergeron in the story, defy the authoritarian state with dance.
Granted vr telemetry is based off of multiple cameras tracking a single individual - still, one wonders how hard it would be to cover a crowd with specific aim of identification and tracking?
For all practical purposes, it is impossible in the current state of security/surveillance video footage.
The example described in the paper in your link is using cameras that are setup to provide very high resolution images of people. It would be like using full-frame portrait images for face recognition, and then expecting that to translate to real-world scenarios where you might only have 20 pixels on a face, and the person is off-axis to the camera.
Gait detection has been discussed for a while, and may definitely be a thing one day, but right now we are barely at the point where pose estimation is a thing in security video. Very far from being able to do high precision pose recognition and sampling over successive frames to model something that would qualify as "gait".
Ive often wondered if shirts with Epaulets equipped with LED IR lights might provide a shroud - or a jacket /hat/helmet with an IR array would always be a good thing.
Imagine Epaletts-like-strips of IR/UV LEDs connecting with magnets to any garment.
Or low cost strips of IR LEDs glued to some neodymian magnets which are attachable to camera shrouds.
Else - Epoxy filled paint-ball rounds to be shot at cameras to obscure view, or fast-dry epoxy rounds for shooting at drones/kill-bot joints.
If you havea BostonDynamics B.I.T.C.H. (Battle Intelligent Tactical Canine Hunter) after you - shoot its joints with epoxy that freezes its legs/joints/sensors)
EDIT: - This is what every Major, Col, Gen should have in their leafs and stars:
>>"Ive often wondered if shirts with Epaulets equipped with LED IR lights might provide a shroud"
Every general/col/maj should have this technology integrated in uniform.
Yes. The hard part is having an "R2" which constantly searches from a back-pack with a painter scattering for cams which can then take offense/deffense that sees said cams and either paints with laser or chem-a-lized paint gun pellets with aluminum paste and epoxy mix... disabling both movement and measure.
You dont wear these in combat (under certain circumstances) (( you develop different frequencies/delivery mthods, but I think your point is more valid.
Yes you can thwart via, but then - you wind up just tracking anyone with it.
Thus you need a "rain" of micro particles in a crowd where every single particle sets off said "anti-sensors"
How cyberpunk do you want to get with the designs?
This creates an anomaly on any surveillance to the point you will get exact time / date and you will be able to get people that can be asked what you look like.
It’s like a padlock. Won’t keep people who try hard enough out, but will keep most modest efforts out. Also, this would prevent a detection from occurring at all, so there wouldn’t be an anomaly any more than a trash can not being detected as a human would.
Dark sunglasses, a baseball cap and an N95 mask work better than flashing a huge light on video surveillance and also deterring people to remember a face and doing it this way gives no date and time information.
Maybe I'm missing the point since both deter automated face recognition but then you could add to the classifier this huge light source as another thing to recognize.
The strobing seems to defeat the effect -- you only need one good frame with a face showing to defeat the hoodie, and this seemed slow enough that I could pick up the face of the demo guy with my own eyes in realtime.
Seems like the strobing is intended to defeat the autoexposure compensation. Otherwise the camera might see the brightlight and reduce exposure to compensate. Concentrating the lighting on the face only and staying far away from cameras might help that without requiring the strobe.
The leds are placed on the chest -- seems like a miss to not place them also around the hoodie near the face. I would suspect even without strobing the over saturation would work should those leds be surrounding the face.
All of that said, wearing something like this out (unless it becomes super commonplace lol) just screams watch me closely (and easily follow me back to some known origin).
> The leds are placed on the chest -- seems like a miss to not place them also around the hoodie near the face.
Based on the documentation, all that would probably do is allow the effect to work at slightly closer distances:
> In addition, the LEDs need to be a minimum distance from the camera to be effective, as the cone of light from them needs to be wide enough to overlap with the cameras view of the wearers head. In practice, this is about 12ft (~3m).
Putting them on the chest would help with ergonomics. It would work even if you have the hood off and the LEDs would be more consistently positioned.
FR developer here: that strobing is very smart, as the mechanical mechanism starts to adjust the strobe generates another adjustment. If the strobing is at a random interval, it becomes difficult to impossible to compensate in software. FR does not work without face images.
I would think randomizing the strobing would be effective as well. All of this defeating of auto exposure adjustments makes me think back to the days of Macrovision and how easily it could be defeated as it was not randomized at all. Then again, it was also easily just stripped out of the signal.
With random timing, you're more likely to get one period that's long enough for the auto-exposure to correct it and see the face. Autoexposure takes time because it has to use a feedback loop to search for the correct exposure, so if the brightness changes fast enough, it can't keep up.
you set a maximum on period so that doesn't happen. if you develop something with a known pattern, that can be programed in to be countered. hence my example of Macrovision. randomizing the values means it is much harder to counter.
That's what they say, but how much autofocusing is actually happening? In the example video the scene is mostly in focus. My guess is it actually messes with the exposure rather than focus.
Agreed that was not a good demo. I assumed that segment was not taken with a cheap webcam/security camera - which is why the auto-exposure worked a lot faster.
At a greater distance with lower MP security cameras, this thing might still work.
I'm surprised nobody is producing swag with the EURion constellation[0] on it. You'd be surprised how many random image processing libraries it breaks.
Or be a foreigner here using an ESTA. Generally committing crimes after giving the government every little piece of information about you is a bad idea.
These seem much less functional and more of a fashion statement. Defeating ALPR readers when you're not a car seems bizarre. Where the OP is actually actively protecting one's identity from being followed.
So even though the name says adversarial, how is it actually being adversarial?
"The patterns on the goods in this shop are designed to trigger Automated License Plate Readers, injecting junk data in to the systems used by the State and its contractors to monitor and track civilians and their locations."
They have mobile ALPR cameras mounted onto cars aimed at where one would suspect a license plate would be visible from that position. So, unless someone wearing a shirt like this is laying in the street, not likely.
Maybe places like parking garages would have some mounted where a person wearing one of these might walk by, but the significantly small numbers would make any junk data insignificant. So, I cannot imagine these being useful to that end. Other cameras like traffic cameras or toll boths or similar mounting locations would also not be hampered by these shirts unless you make them into a complete body wrap for the car.
However, I do see them being useful as conversation starters when someone asks WTF you're wearing. It would have a much better chance of increasing awareness of ALPR tech and possibly other camera tracking tech in general. But even that is pretty much meaningless in the grand scheme of things.
This would not be highly effective with most modern security cameras.
First, color low-light video is becoming more common, as image sensor technology has progressed. When a digital/surveillance camera is in color mode, there is an IR filter in front of the lens to remove these wavelengths from reaching the imager (they cause color tinge issues).
Even for cameras that are not in color mode in low-light video, most newer units have good dynamic range, whereas these low-power IR LEDs would likely not be able to fully obscure a persons face in the video.
Not every color camera has an IR filter, many phones and webcam lack such a filter and can see IR very well (on my phone it has a distinctive purple hue).
Probably this tint is compensated for in normal light.
That distinctive purple hue is a result of the IR filter. Filters on consumer cameras don't block 100% of IR, which is visible as that purple color. Useful for checking if remotes are working properly.
I don't think it's because of the IR filter. I have removed filters from several color webcams and they always see IR as purpley. I think it's the bayer filter - the different color channels clearly have different sensitivities to IR.
Yeah, what I mean is the existence, the visibility of a slight purple hue. If the camera has no IR filter, you don't just see a slight purple hue on the LED itself, but you see the light just like you'd see a visible wavelength LED.
I actually was obsessed with IR cameras and mods for some reason around 10-12 years ago. I remember using developed empty photo films to block visible light and let only IR through after removing the IR filter.
If you look at a quantum efficiency chart you will find that CMOS sensors are actually sensitive to 850nm IR light almost identically at all photosites regardless of the colour filter and then they gradually decline together as the wavelength increases. The colour cast is mostly due to the tuning and white balance of the image.
At least in China face recognition software is good enough that they can recognize people even when wearing covid mask. I mean even iphone face id works with masks, scarfs, sunglasses and in the dark (i didn't test all at same time so maybe there is hope...). So if you want to be anonymous then you can't, sorry, thats the world we live in currently.
IF you submit a UK GDPR DSAR or equivalent in your own country, the cops will deliberately not hand over the data, including Axiom bodycam footage, under the pretence its all used for law enforcement purposes, which the legislation lets them do. I know, I've tried. I've even been told its up to the individual officer whether they want to submit their bodycam footage, which is a great way for the police to cover up crimes, considering they already have the option to switch on the recording, and these bodycams have a 30sec buffer which gets included when they enable recording. Police cars also have built in AV recording to monitor what goes on inside a police car.
You would be amazed at how criminal the Police in the UK really are, they do a good job of PR on social media and the press doing them favours to get leads on stories.
You are better off having your own stealth AV devices recording your every move and sound 24/7, built into jewellery, head torches, clothing, dash cams for vehicles, and smart phones recording all the time, so you dont even have to bother taking a photo or video, just point and record. And where possible have it streamed live back to your servers with a dead man's switch.
Its the only way to combat the intelligent entity's who seek to dominate.
It's a crime in the country where I live, article 138.1 of the Criminal Code. Illegal manufacturing, purchase or sale of special technical means intended for covert acquisition of information.
It's even more fun when the law explicitly permits[0] recording of the violation of the law but police and judges would do anything (and succeed!) declaring what that time it wasn't permitted or... wasn't violation of the law, so you couldn't record.
[0] "doesn't forbid" would be more accurate description
> the wyze cameras that make my own surveillance around my home
They are too obvious, if you have a brick property, put a few bee/insect bricks at different heights into the walls and then drill through from the inside of the property and put a slim line camera like the AXIS P12 Modular Camera Series in one of the larger bee/insect holes. Doesnt matter then if the individual has their face facing the ground, which is lot of the time, you'll still get a good image of their face.
You can find even cheaper cam's if you search aliexpress.com.
Our gated community spent $80,000 on upgrading their cameras system and bought state-of-the-art 5MP surveillance cams (total of 5 of them) that "lets you read license plates from within 100 meters" and "does a color-in-the-night type recording". They were sold on upgrading the whole freaking network of cables to cat 9 for God only knows reason. And I kid you not - less than 2 months later, we were hit and 6 cars were stolen. Apparently it doesn't matter how fancy your camera is, if the perpetrator setup a stand alone red laser at cost of $6 that shines directly at it...
Reminds me of the time I was looking at my uncle's AV setup. Back then, you had multiple ways of delivering video between components: RCA, Composite and a variety of others.
The tech person had every possible path wired up with Monster video cables. A waste of probably $500!
For your gated community, it seems like a detection for the video feed going haywire that calls the cops might be useful.
So all those digital video cameras that are pointing around to be used in visible light or low light conditions (like those for retail surveillance) will now have a person with a blinking light on them as people look at the closed circuit feeds during regular business hours.
I wonder if such strong IR light is capable of harming the eye, even if we're unable to see it.
High power LEDs in the visible spectrum are terribly bright, and when I accidentally get them shined into the eye, it really feels concerning if any harm has been done.
"If you buy the largest IR floodlight available on Amazon and you stare at it 200 mm (~7") from your face for more than 1000 seconds (~ 16 minutes) there is a possibility you might damage your cornea. So don’t do that. And I doubt anyone would do that because these things get pretty hot at that distance."
Near-IR aiming lasers (such as the AN/PEQ-2 and AN/PAQ-4) can indeed cause permanent damage, among USAF CATM armorers there was a rumor of at least one idiot burning his retina trying to "see" the beam directly.
We can stare at a burning campfire and barely get damaged. Long term exposure is another thing. The emitter power would have to be insanely high, or some kind of laser.
Yes, but a campfire emits a lot of visible light, so our eyes adjust accordingly. While we're not cats with their amazing pupil dilation, we have some too. If one wears an IR-shiny hood, they may be at higher risk in the low-light conditions, if their eyes adjust to the dark, brain not realizing they're flooded with IR.
Retinal damage is not the only kind of damage the eye can be subjected to. The cornea has no blood flow to cool itself and is fully exposed to incoming radiation.
A campfire has radiance limited by its temperature. It can put out a lot of power, but that power can't be focused to a high energy density at your retina. An LED can be nearly a point source, and can deposit a lot of power at a small spot, i.e., "almost a laser." Moreover, being invisible, an IR LED doesn't cause your pupil to close. And LEDs have gotten powerful enough to be dangerous. The regulations for optical radiation exposure now cover LEDs as well as lasers.
Gait analysis is a more surveillable fingerprint anyway, since it can work with lower resolution cameras. Tag someone on a high res camera with facial recognition and as long as you've seen their gait once, you'll be able to identify them again.
Has gait analysis been studied by people other than the ones selling gait monitoring equipment? It smells to me like bite mark analysis, blood splatter analysis, polygraphs, and other "miracle" technologies/techniques that turned out to be horsecrap.
There are people I know that I can recognize by the way they walk, so the idea is at least plausible. I think the main problem is/will be from using the analysis inappropriately.
It's one thing to narrow down a city-sized population to a list of 50 people known to walk with a similar gait, and then follow up with other forms of investigation before drawing conclusions. It's another for a cop to look at the top match and go arrest that guy, as they've done more than once with facial recognition.
I don't think gait analysis would hold itself through a trial or be bought by a jury. Maybe to correlate different videos of possibly different suspects? But if you are getting regularly videoed by security cameras while circumventing rules you are probably already wearing different shoes or pants and that can go a long way to defeat gait analysis algos.
It's quite literally more accurate than fingerprint or blood splatter analysis and those send people to prison on a daily basis - despite being little more than snake oil techniques.
The person knows no cameras exist doing gait detection, but wants one. They make a random comment on HN, a forum known to be read by people that is full of people will take an idea to completion "over a weekend", as a challenge to get someone to do something for $1000. That's a ridiculously low amount for something to come into existence that would serve their need directly. Hell, there's probably people here that would do something like this for the clout.
No, I know that where security cameras are positioned and the amount of processing they currently have there will be no gait analysis being done on the camera itself. They have enough trouble telling one person from another reliably.
I was being facetious as I don't believe hackernews readership is that enterprising.
There have been pretty huge leaps over partially obstructed facial recognition algorithms in the last few years -- I think state of the art is approaching no meaningful loss of match mask vs no mask.
iPhone face unlock has worked while wearing a mask for a while now.
You know how you can identify exactly which family member is walking in the door from across the house, and maybe even what mood they're in? Or how little clarity you need to identify your child or spouse on a 480p camera feed? That's what machine learning makes possible across all types of sensor input, picking out those little but distinct patterns. There's really no way to be anonymous in public once ML surveillance software is widespread.
This is not my understanding, I have read at least 7 or 8 papers that seem to have for various models and techniques reduced the delta between masked and unmasked recognition to be very similar on false positive and positive rates.
That said I have no insight as to how many of these techniques have been found to scale well or have started to make it into product. It has been publicly reported that NEC’s NeoFace (a system that many police and govt use) newer versions does indeed have occlusion (mask) recognition operating at very high levels.
anyways thats just my understanding as an interested bystander -- not in the field.
No, older smart phone cameras lacked IR filters because the IR filter didn't only block IR light, but also a distribution of wavelengths around IR. Filtering out IR means many of the visible light wavelengths are attenuated as well, decreasing the amount of light arriving at the sensor.
With better noise reduction algorithms, more sensitive sensors and lower noise sensors, IR filters are now almost always used in smart phone cameras.
Also, without an IR filter those cameras could partially see through light clothing in bright conditions. This of course formed a creepy online community and enough backlash that the manufacturers decided to include IR filters for all future products.
I work with IR sensors at work and modern smartphones pick up NIR frequencies as well. It's picked up by the red channel so combining that with the blue is probably what's making the purple color.
My guess would be that something bright enough to wash out the red channel would still leave the blue and green intact. So there should be enough information that the image is degraded but not necessarily blinded.
But I could be wrong, depending on how independently the color sensors and processing work.
> As the hoodie uses IR light, it’s effects are imperceptible by human eyes when activated, only effecting IR sensitive equipment
Just because you can’t see it doesn’t mean it can’t hurt you. If it’s it’s not directed away from your eyes, a bright IR light an inch away from your eye can cause harm.
Now that more and more cameras use AI models to detect and infer things in real time, wonder if a very specific adversarial t-shirt could render the wearer completely invisible to the AI (not to humans looking at the CCTV monitors or the video recordings). I think William Gibson wrote something like that in the Zero History book, "the ugliest t-shirt in the world" or something
There are a bunch of papers about adversarial patterns on shirts, glasses frames, make-up, etc. The problem is that the more discreet it is, the more overtuned to specific models it'll tend to be. If you really want to reliably foil face recognition, you can walk around in a ghillie suit and camo face mask, I suppose.
interesting. i always thought that IR LEDs wouldn't be bright enough, guess they are (at least in total darkness)!
as mentioned all over in the thread, adding jitter would probably help.
worth noting that every camera is going to have a different AE algo, so it could not be relied on and you only need one frame to succeed for the system to be defeated.
what you would definitely get, however, is excessive triggering of any region of interest detection for the video stream. not exactly low key!
all told, pretty cool experiment nonetheless!
on a related note, i often used to think it would be cool to put IR LEDs in public places where people take pictures, sort of like hidden messages or art for people to discover, but only after they look through their phone cameras.
Wonder if CCTV cameras will start to detect incoming infrared and use a visible light spotlight in response.
>Darkness is no longer the refuge it used to be
Tbh from personal experience darkness still is the refuge for people like me going on quiet night walks, until the time I got mugged. Darkness was a great refuge for them, not so much for me after that day. Or the time that a burglar entered my house at 4am and was cleaning out my room, reaching over my sleeping body to grab my phone charging next to my pillow. Great refuge, felt so safe.
For home surveillance I may just have to equip motion spotlights as well, now.
OK, but why? If you aren't doing anything illicit wearing this seems like a hassle (does this even work on non-IR cameras?) and if you are doing something illicit wearing this will make you standout.
Curiously enough, the 940nm NIR wavelength used in this hoodie was recently shown to reduce covid symptom severity resulting in an average 3.7 day earlier hospital discharge when compared to a placebo group using the same LED vest without turning it on.
So this hoodie is actually a covid hoodie too (assuming you turn it inside out)
As someone with experience in this area I always find these proposals to add IREDs to block identification hilarious.
First, almost every security camera on the market these days has a IR cut filter switcher which means these aren't going to be affective against cameras in the daytime as the IR light is filtered out.
Second, there are other methods of identifying people using cameras if facial recognition fails. You can identify people by their gait, or by their clothing, track them back to their car and get their license plate, etc.
Wouldn't IR filters defeat this once it becomes an issue? You can just glue them on the front of the lens or it can be on the sensor though the latter requires a new sensor.
Do you have a source or something? I'm not trying to doubt you but I'm interested in learning further. I work on cameras with IR emitter and there are many safety protocols, emitter covers, warning signs, etc. for a simple IR camera that we are instructed to abide by.
I work with a company that makes high-power IR illuminators and we just went through certification testing with some units that are roughly 100x the output power of this example and they did not pose an eye hazard.
There is a bunch of stuff online about exposure to IR-A wavelengths, which covers the common IR LED/illuminator wavelengths. You certainly CAN build something which can cause various kinds of eye damage, but it is going to be a unit that you can literally feel the heat off of. For practical purposes, you couldn't make a wearable battery powered device that would be likely to cause eye damage without the entire thing being very large and unwieldy, assuming of course you are using standard incoherent (eg:non-laser) source.
The actually invisible NIR is 940nm but generally IR therapy devices come with blockout goggles like you'd use in a tanning bed.
I'm not sure the power levels of this hoodie but if you're 3 meters away from an LED it's way different from if you are 10cm away.. I wouldn't assume safety when making up new use cases for a 'safe' product.
Privacy is both important and valuable. However, during the product demonstration, was it truly necessary to have the actor dressed entirely in black, walking into someone's backyard? What specific use case do they believe they are showcasing with this demonstration?
In last few seconds of video (around 0:49) you can clearly see the face even with strobe LEDs turned on, if CCTV looks directly at the face and not from the angle, so they actually showed it doesn't really work with well placed CCTV at face level.
The way the LEDs are blinking together one can definitely see the face in the periods they are all turned off. The only advantage I see is that this is not humanly possible to do in real time, but not sure how big an advantage this is.
Hasn't this already been done before with hats, hoodies, and glasses? What makes this any different from projects 10+ years ago that used IR LEDs for this purpose? The model of the LED? Strobing effect?
Could these IR-LEDs be effective for movie theaters to prevent pirate cammers from creating viable recordings? Seems like it'd be pretty easy to implement if the studios really felt threatened by piracy.
The idea is the IR lighting would be placed up front beside, below, or above the projector screen, and interfere with the visual recording device to ruin the captures. In my experience digital camera sensors are sensitive to IR light in low light environments.
I've never tried in a movie theater :p but have noticed this effect by pointing my phone camera at an IR television remote and pressing some buttons. The IR LED lights up and is only visible to the camera and not the naked eye.
Anyhow, I'm more just curious why studios / cinemas don't take relatively simple measures like this to prevent pirate recordings from being created. Perhaps the IR would be too easy to circumvent via a physical lens filter.
During daytime, I assume cameras use visible ambient light (that's at least how my baby monitor works fwiw :-). And of course any regular camera e.g. Dslr I assume wouldn't care much?
I don't like the idea of a hoodie blasting high power IR which could easily get into your eyes. Perhaps it'd be better to make a camera-shy baseball cap.
if you really wanted this to work you would wear the strobes all over so they don't pick up arm movements or leg gait, i'd even go as far as all of your knees, feet, elbows, shoulders, and wrists/fingers
if you can eliminate your movements and shape then you're ahead of the game, otherwise you're identifiable
Does it have A 100% FARADAY cage pocket to put the biometronomic tracking device (phone) within in addition to shield that, plus an array of UV lights to shim cams off?
How many people who are happy to see your face are watching you on surveillance cameras? It's not like this hides your face from a human, and you can always put the hood down if you're at your friend's house or somewhere else you trust.
you can't skulk through a crowd worrying about surveillance cameras without... worrying. I'm saying that selecting the word adversarial to describe this garment is not a coincidence, it's an attitude, and I'm linking it to the emotional life of people who would be interested in the garment.
worrying is stressful anxiety that your friends (who can see your face) can't completely erase for you. adversarial essentially means "looking for anxiety". I'm saying step back, it's not the only way to live.
downvoting btw is also adversarial so have at it, you'll feel better... not :)
Or it's something people enjoy as a thought experiment? I work adjacent to computer security industries and have always found the sort of cat-and-mouse of circumventing various detection and security mechanisms to be engaging and interesting.
There's certainly people who are doing it out of anxiety, either intentionally or because of their own history or mental issues. For them, it may not be easy or possible to just decide to live another way. Either there actually is something they need to be adversarial towards or they're well aware that it's not necessarily rational, but it's not any easier for them to stop than it is for a depressed person to "just be happy".
> who wants to wear an adversarial garment? adversarial people.
I disagree.
Who wants to wear an adversarial garment? People stuck caught under the spying eyes of adversarial governments and adversarial surveillance capitalism companies.
The objective of the hoodie is to prevent face recognition.
Those who have come down with switcheroo fever are saying "aha! This actually makes you more recognizable, not less!"
But this is not a trick question and this is not the type of thing where the switcheroo is meaningfully responsive.
The objective is to prevent a camera from recognizing your face, and it does that. If you have tested positive for switcheroo fever, you should start your comment by stating out loud that you understand this is the objective and you understand that it does indeed achieve the objective before proceeding elaborate on your own personal variation of the switcheroo. And perhaps, if you dare, proceed to challenge mode, where you acknowledge that this might be a trade-off that was consciously chosen, and does not come as a surprise to the people who designed the product.