AVG Announces Invisibility Glasses | Hacker News Reader

AVG Announces Invisibility Glasses

now.avg.com

57 pointskrnaveen43 comments

Speaking of IR... Not about privacy, but DIY medical stuff.

My wife suffers from restless legs syndrome. Pretty much, whenever she goes to bed, her legs feel as if they have ants crawling on and inside them. I know that I can use massage (she taught me, she's an MT) to reduce the restless feelings, but it's just a stop-gap. What causes it?

Turns out, its some -NO precursor that appears to cause this. Why does this precursor build up? I have no clue. But getting rid of it does alleviate the pain.

How I found to get rid of it, a medical paper said that NIR light is found to penetrate in the skin and is enough energy in the photons to break up the precursor. The medical paper was using a grid of low power IR lasers, with a total of 3W dispersed over a 12inx6in rectangle.

I made a 30 IR led (matched the frequency of IR from the paper) that outputted 3W of light. I use it on her when her legs acts up, and it settles them down. Usage over a week gives her days of rest from RLS.

Do you have any details on this build? My father has RLS and nothing seems to work for him. A quick google shows me you can buy the units used in the study, but they are quite pricey.

Sure do.

Scientific article I based my device on: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2936319/

"NIR has a wavelength of 880 nm to 890 nm"

"Researchers hypothesize that the success of NIR treatment lies in its ability to increase bioavailability of nitric oxide (NO) in the lumen"

"890 nm wavelength light, pulsed at 292 times/s, with a power output of 600 mW/cm2"

"The mechanisms with which NIR can alleviate RLS symptoms are not clear. One supposition can be made: light has been shown to generate NO in the endothelium, which through a cascade of events leads to vasodilation. Vasodilation is also the result of exercise, one of the few non-drug related treatment options that decreases RLS symptoms."

The above is the basis of my research. From this, I can make a device that does roughly this.

_____________________________________

My Bill of Materials are the following:

     12v @500mA wall wart
     30 http://www.digikey.com/product-detail/en/TSHF5410/751-1211-ND/1681346  IR LEDs
     barrel jack to fit wall wart
     perf board to fit LEDs
     project box
     IR-transparent acrylic (window in project box)

I used a scrounged wall wart (free), the LEDs are $20, the barrel jack was scrounged (Free), the perf board was $5, the project box was $9, and the acrylic was scrounged (Free). Total cost for non-scrounged was $34. I'm also part of a hackerspace, so scrounging was easy for me. I also used a bit of hot glue to make smooth interface where the skin would touch.

The circuit design is here: http://imgur.com/5Op2ndU

Solder them together through perfboard, and put in project box. Cut out a window on the box and put in the acrylic window. Now, all you need to is plug in the barrel adapter to the wall wart, and it works.

This is what I use. It puts out roughly 1.35W of IR light. I misremembered about using 3W. But this works, as it's a reproduction of my build.

Very very interesting. I have a whole family that suffers from the same, five siblings. Now we need to try that. Do you have a photograph of the contraption you made?

I'd be glad to. It's at home at the moment. I'll probably, with my wife's permission, send it up to my github.

Also, feel free to contact me. https://keybase.io/jwconway

They're only 7 years late.

https://www.schneier.com/blog/archives/2008/07/sunglasses_th...

Its my understanding that face recognition is tough to crack on a mass surveilance level. People wear hats, glasses, sunglasses, scarves, etc on top of things like different hair styles, different glasses frames, etc. CCTV quality isn't good enough to identify faces and even high quality video can be tricky if you don't get enough high quality frames of the person's face looking straight at the camera.

It looks like identifying people via their gait was a bit of a research hurdle not too long ago and there are a lot of articles from around 2012 that claimed this problem has been solved. The idea was low quality CCTV footage really can't be used to identify faces, but can be used for gait analysis. I wouldn't be surprised if there was a signature gait database somewhere in use by intelligence and law enforcement services. The same way the Stasi kept a database of people's scents in jars.

Now with everyone carrying a cell phone, I imagine immediate identification is easier than ever. IMSI-catcher devices like the Stingray are trivial to implement. Phone carriers easily give up information about their customers, including GPS history, sometimes even without a warrant. License plate tracking is common as camera deployment and plate OCR is very affordable for many municipalities. Tracking via toll system RFID transponders is common as well.

Not sure who these glasses are aimed at. It looks like anti-security theater designed to get press and lighten wallets.

Not a bad start, but hopefully something better comes later. This would just be an arms race scenario- they'd only have to develop algorithms to work around this, and worst of all, they still have your actual picture on-disk so they have all the time they desire to make it work.

I seem to remember an anti-camera system deployed at movie theaters years ago which would look for camera sensors, and then blind them with IR lasers. While probably not as portable, something precise rather than the shotgun approach they're using here might work better. Bonus: you don't need to have it mounted to your face, either- it could be attached to your hat, shoulder, chest, etc.

> anti-camera system deployed at movie theaters years ago which would look for camera sensors, and then blind them with IR lasers

I think it is unlikely. IR lasers can still damage eye sight.

Looks like shooting IR at the audience is how some/most of these systems work:

http://archive.wired.com/entertainment/music/news/2004/11/65...

but yes, it does sound like its probably like plain old IR light and not a laser.

Wouldn't this all be undone by an ~$50 IR filter attachment on the camera[1]? No need for new algorithms or computing power to fight the ultra-bright sections, just limit the camera to the visible spectrum and the "anon light" is defeated.

[1]http://www.bhphotovideo.com/bnh/controller/home?O=&sku=23078...

Depends on if it's using IR to illuminate. I guess I was thinking of dystopian settings where there's a camera on every signpost, tracking your every move. In these cases, I'm assuming they'd be very interested in what goes on at night.

Now, realistically (at least for now) the bigger concern will probably be cameras tracking you in well lit malls, stores, things like that. In those cases, you're probably very right. Either my idea wouldn't work in those cases, or something in the visible spectrum would have to be used instead- something that couldn't be filtered out.

Doesn't defeat the face detection in libccv:

http://imgur.com/HNPBNTJ

http://imgur.com/P3XnlwK

Edit - I don't mean to kill interesting conversations around this, but be aware we're looking at an implementation that simply doesn't work, so we should only really be talking about similar but different possible approaches. Anything like this is so fundamentally flawed as to not even currently work against currently used algorithms.

So if we want to go down that road, what more extreme things might we need to do?

The article does not claim that the glasses prevent face detection. It claims that they prevent face recognition. These two concepts are different.

Face detection - determine the location of a face or faces in an image.

Face recognition - given a location of a face in an image, correctly predict who's face it is.

Check out the OpenCV Facial Recognition Tutorial for more info: http://docs.opencv.org/modules/contrib/doc/facerec/facerec_t...

> The article does not claim that the glasses prevent face detection.

Their first image seems to be showing the difference between detecting and not detecting a face ("Facebook detects a face" and "No face detected").

Comments like this, too:

"They claim to break face detection when the lights are on."

So I assumed they were claiming it breaks face detection.

This defeats regular face detection but you are now literally wearing a big, glowing sign that says "this individual is hiding his/her face!".

Like many privacy tools, this would work best if everyone was wearing them.

It'd be more effective for everyone to vote for people that intend to break down and illegalize all of these facial recognition things, if we're all going to organise something anyway.

That's a bit tougher though.

For one, privacy concerns are just one very small part of what people may be looking for in an elected official. So small that most don't even have an official stance on the matter, aside from canned politician spin like "I value the privacy of all citizens" that actually means nothing. On average, I'd say that most people would still rate economic and diplomatic policies much higher.

Like just about everything else, this boils down to an education problem. Not just for the voters, mind you- it's even more important to educate our elected officials and judges. This way, they'll be able to make rational, informed decisions when the NSA comes asking for more money / power. (This is of course assuming the NSA doesn't have huge caches of blackmail evidence on the officials making the decisions...)

Or the algorithm that found your face will just learn to find these.

> This defeats regular face detection

It doesn't even do that.

http://imgur.com/HNPBNTJ

http://imgur.com/P3XnlwK

except if it's done with a program it wouldn't be able to tell

I thought April Fools arrived early this year...

I agree. If it's not a joke, they should have tested those pictures on Facebook's facial recognition system.

I strongly suspect even a simple system will easily still recognize that guy.

I think they tested the first one, and facebook failed, although both approaches have no effect against the libccv face detector

http://imgur.com/HNPBNTJ

http://imgur.com/P3XnlwK

Whatever dude, these are the best invention...

...to use with my glow sticks when I'm boogie skating..

Although it is a good start, I wonder if the effect of IR LEDs can be solved by filtering? I mean, can someone develop a firmware or use color filters to filter out infrared light?

I think the filtering may be done using better Bayer filters to filter out infrared, but I know almost nothing about digital photography.

Yes, almost all cameras above the dollar range will include filtering for this frequency. You can find out by pointing a TV remote at your phone camera and seeing what the impact is. My iPhone can't see the IR emission at all with it's rear camera, the front camera doesn't have a filter and gets absolutely blasted. It would be devastatingly effective against monochrome CCD cameras, where the same LEDs as used in the article are used for invisible illumination of a scene.

https://i.imgur.com/zkI5p0B.jpg

https://i.imgur.com/BfRWTbq.jpg

If I remember correctly, a cut piece of floppy disk does a reasonable job of being a blocking filter for nearly everything that's not near IR, though that's less of a handy thing to know these days.

The IR nut is tough to crack if your camera does night vision. Those cameras with a ring of LEDs around them? They're IR transmitters. The camera is designed to detect IR, so it can't filter it. This is a popular method of seeing in the dark and is used on systems way more expensive than dollar store close-outs.

Would it be that difficult though to have a modified Bayer filter with an RGBI array instead of RGBG?

You would have to use different dyes for all pixels, since the currently used ones let IR light pass, but it doesn't sound that difficult to do (since we already have both IR-pass and IR-block filters).

> Yes, almost all cameras above the dollar range will include filtering for this frequency.

Unless they are security cameras and use IR lights to illuminate scene when it's dark.

Are they trolling? Those are just some glasses with ir light and ir reflective coating. These will make your face more visible to cameras in the dark. Also facial recognition algorithms can easily be adapted to work around the stupid glow. They only need like 10% of your face anyway.

This is just a 1% solution, which may be enough to defeat mass data collection as long as it remains niche. For a bigger effect, carry light strong enough to saturate even a HDR camera.

If you remember the old (VHS) Macrovision, it used a pulsing contrast to mess up automatic gain control circuits. A good way to defeat this is by putting a POT on the AGC so you can tune it manually.

A way to apply this to the glasses: put a IR strobe on. Pulse it at .5s, with a .05s IR led on time.

You should be able to get away with this using an array of IRleds, battery, and a 555 timer.

This is clearly a concept and not a production product — probably more about encouraging discussion regarding privacy in public than about executing on the things it talks about.

I lived in a city with red light cameras, and I considered building a license plate cover with the same principles as a tiny form of non-violent protest.

Do you mean "red light camera" or "license plate scanner" (scanning everybodies license plate)? I don't see why a red light camera is something to protest about.

Here in Brazil they are common, and I protest by not running red lights. Works every time!

'Do you have anything to hide?'

I don't think is much new, they've been made before...

https://www.schneier.com/blog/archives/2008/07/sunglasses_th...

Although I think it was meant mainly as a way to prevent you getting recorded on CCTV.

Surgical mask seems like more effective.

That and a pair of sunglasses

If you wear a pair of those glasses, everyone is going to know who you are.

Reply on news.ycombinator.com