Stories
Slash Boxes
Comments

SoylentNews is people

posted by janrinok on Sunday October 13 2019, @05:38AM   Printer-friendly
from the as-if-Big-Brother-needed-more-bandwidth dept.

Watch Out, MIT’s New AI Model Knows What You’re Doing Behind That Wall:

For better or worse, AI can now figure out what you're doing even without "seeing" you. The MIT Computer Science & AI Lab (CSAIL) has unveiled a neural network model that can detect human actions through walls or in extremely dark places.

Although automating the process of action recognition from visual data has been a computer vision research focus for some time, previous camera-based approaches — much like human eyes — could only sense visible light and were largely limited by occlusions. The MIT CSAIL researchers overcame those challenges by using radio signals in the WiFi frequencies, which can penetrate occlusions.

Their "RF-Action" AI model is an end-to-end deep neural network that recognizes human actions from wireless signals. The model uses radio frequency (RF) signals as input, generates 3D human "skeletons" as an intermediate representation, and can track and recognize actions and interactions of multiple people. The skeleton step enables the model to learn not only from RF-based datasets, but also from existing vision-based datasets.

Researchers say RF-Action is the first model to use radio signals for skeleton-based action recognition. "There are lots of potential applications regarding human behavior understanding and smart homes. For example, monitoring the elderly's abnormal behaviors such as falling down at home, monitoring whether patients take their medicine appropriately, or remote control of smart home devices by actions," says the paper's co-first author Tianhong Li.

Using RF in the "WiFi" bands. 25 hours of data was all it took (or all they collected) to train and test the AI. This article was unclear if the WiFi RF used was active, or passive although earlier reporting specifically mentioned passive.

MIT CSAIL RF Action site has a link to the paper:

This looks like an update of the story we first published in 2015, but now including AI.

Previously:


Original Submission

Related Stories

See Through Walls by Passive Radiation From Wi-Fi 19 comments

Researchers at University College London (UCL) have devised a system for detecting the Doppler shifts of ubiquitous Wi-Fi and mobile telephone signals to "see" people moving, even behind masonry walls 25 centimeters thick. The method, which could be useful in situations from hostage-takings to traffic control, won the Engineering Impact Award in the RF and Communications category at this National Instrument's NI Week 2015 meeting (which convened in Austin, Tex., 3-9 August).

Other researchers—notably Dina Katabi and Fadel Adib of MIT—have built through-wall radars in the household communication bands, but these are active radars that transmit as well as receive. The UCL technique uses only passive radiation—from Wi-Fi routers (using emissions in any of the IEEE 802.11 b, g, n, ac), ambient GSM and LTE mobile signals, and other sources—so there is nothing to betray the surveillance. The system calculates the positions of hidden target by comparing two signals: a reference channel, receiving the baseline signal from the Wi-Fi access point or other RF source, and a surveillance channel, which picks up Doppler-shifted waves reflecting from the moving subject.

Tan and company built their "high Doppler resolution passive Wi-Fi radar" on two multi-frequency, software-defined, FPGA-based transceivers (National Instruments' USRP, or Universal Software Radio Peripheral. The system compares the reference and surveillance signals, interprets the very small frequency shifts, and reveals the hidden subject's location and motion.

This article has been visited 15 million times by teenage boys.


Original Submission

X-Ray Vision? Technology Making It a (Sort-of) Reality for $300 9 comments

A group of researchers led by Massachusetts Institute of Technology professor Dina Katabi has developed software that uses variations in radio signals to recognize human silhouettes through walls and track their movements.

Researchers say the technology will be able to help health care providers and families keep closer tabs on toddlers and the elderly, and it could be a new strategic tool for law enforcement and the military.

"Think of it just like cameras, except that it's not a camera," said Fadel Adib, a researcher on the MIT team developing the device.

"It's a sensor that can monitor people and allow you to control devices just by pointing at them," he said.

Work began in 2012 to determine how wireless signals could be used to "see" what's happening in another room, said Katabi, who directs the MIT Wireless Center.

"At first we were just interested ... can you at all use wireless signals to detect what's happening in occluded spaces, behind a wall, couch, something like that," Katabi said.

"It turned out that we were able to detect that. And when we figured out we could detect that, we started asking more advanced questions: Could we use it to detect exactly how people are moving in a space if they are behind a wall?"

Using radio signals to see through walls: Cool. Handing new spying tools to government: Not Cool.


Original Submission

This discussion has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 0) by Anonymous Coward on Sunday October 13 2019, @06:37AM (1 child)

    by Anonymous Coward on Sunday October 13 2019, @06:37AM (#906560)

    Ceiling Cat [knowyourmeme.com] CSAIL AI is watching you masturbate.

    • (Score: 0) by Anonymous Coward on Sunday October 13 2019, @02:19PM

      by Anonymous Coward on Sunday October 13 2019, @02:19PM (#906615)

      You're holding it wrong.

  • (Score: 3, Interesting) by jmichaelhudsondotnet on Sunday October 13 2019, @10:58AM (9 children)

    by jmichaelhudsondotnet (8122) on Sunday October 13 2019, @10:58AM (#906586) Journal

    'Their "RF-Action" AI model is an end-to-end deep neural network that recognizes human actions from wireless signals. The model uses radio frequency (RF) signals as input, generates 3D human "skeletons" '

    I have been wondering if the continuos wave rumors from the snowden docs are real, and why no one besides obvious disinformation agents like mr appelbaum have dared speak about them. (and why are the snowden docs still not available anyway?)

    This leans towards my, til now beyond paranoid, belief that cell towers, wifi and other transmitter/receives could be used to image every room in your house at any time, or 24 hours a day, and that on a massive scale.

    This confirms the tech is plausible(confirming my biases and prejudices regarding such information are actually pretty spot on) and if MIT has it now, it has probably been in play for a while.

    https://listverse.com/2018/08/15/10-reminders-of-the-realities-of-mind-control/ [listverse.com]

    Seems very similar to what this guy John St. Clair Akwei claimed in the 90s(although can still not find any other lengths to these case details), like they are trying to get us used to the idea that what has been going on for 30 years *just might* be possible. Remember if it is, all shared with israel and these people are behind the troomp admin, kushner, etc.

    Until now there have been virtually no public scientific articles about consumer bands of the emf spectrum being used as radar, but this is proof of concept.

    To think the nsa hadnt been researching this for a long while at this point, would be incredibly unintelligent and trustful of people who do not deserve trust. And if it seems like people know when you are leaving the house before you do, and I have seen evidence of this, an AI to recognize when you are putting on your shoes would be pretty handy to help you time your paramilitary street gang to get on the right bus.

    thesesystemsarefailing.net
    decultification.org

    • (Score: 2) by Unixnut on Sunday October 13 2019, @01:06PM (3 children)

      by Unixnut (5779) on Sunday October 13 2019, @01:06PM (#906598)

      > Until now there have been virtually no public scientific articles about consumer bands of the emf spectrum being used as radar, but this is proof of concept.

      https://en.wikipedia.org/wiki/Passive_radar [wikipedia.org]

      I don't know if there have been public scientific articles themselves, but the technology is well known and documented.

      The state of the art is most likely classed as a military secret. However if tinkerers can build their own passive radar systems with (almost) off the shelf parts, I suspect what powerful states and intelligence agencies do is miles ahead. If they can do it at high enough frequencies (e.g. from cellphone towers), then they should be able to track objects inside walls, although they would have to have really good filtering to handle the scatter and reflections.

      DIY Passive radar example:

      https://www.rtl-sdr.com/passive-radar-dual-coherent-channel-rtl-sdr/ [rtl-sdr.com]
      https://www.rtl-sdr.com/rtl-sdr-based-passive-multistatic-radar-used-track-aircraft/comment-page-1/ [rtl-sdr.com]

      • (Score: 2) by JoeMerchant on Sunday October 13 2019, @03:08PM (2 children)

        by JoeMerchant (3937) on Sunday October 13 2019, @03:08PM (#906632)

        The "really good filtering" that impressed me first was for audio: arrays of microphones delay-tuned to superimpose on a particular point in the room, actually really simple and they were doing it in the 1960s to listen in on a table in the middle of a crowded noisy restaurant, 100 microphones laid out around the room with the right delay tuning meant that the table of interest audio signal was amplified 100x larger than the surrounding noise.

        More recently, systems have been built and demonstrated that can locate arrays of microphones on an arena scoreboard in the middle of a 20,000 seat capacity crowd basketball game and "tune in" on any seat in the house, even after the fact - simply record all the microphone channels separately and delay-mix them to listen in on the conversation(s) you want to hear. It's like the kiss-cam, but for audio.

        If it works for audio, all you need is higher bandwidth, higher frequency remixing and it should work for 2.4GHz signals as well, and it's not like there's a shortage of cell towers - particularly in urban areas.

        --
        🌻🌻 [google.com]
        • (Score: 2) by opinionated_science on Sunday October 13 2019, @03:26PM (1 child)

          by opinionated_science (4031) on Sunday October 13 2019, @03:26PM (#906637)

          The maths for filtering is very well understood and modern machine learning is making leaps and bounds in build detection networks.

          I have a FLIR camera, and you can *clearly* see human reflections and impressions of electric cables within walls...

          • (Score: 2) by JoeMerchant on Sunday October 13 2019, @04:32PM

            by JoeMerchant (3937) on Sunday October 13 2019, @04:32PM (#906650)

            The other thing that has made leaps and bounds in the last 30 years is high frequency digital signal processing. In the 1980s, audio (to 50KHz) was still somewhat challenging - 44.1KHz 16 bit stereo was actually pushing the envelope a bit when audio CDs were first being designed. Today they're processing 10s of MHz signals like they used to do audio.

            --
            🌻🌻 [google.com]
    • (Score: 5, Informative) by janrinok on Sunday October 13 2019, @02:02PM (4 children)

      by janrinok (52) Subscriber Badge on Sunday October 13 2019, @02:02PM (#906609) Journal

      Passive radar has been around for decades, and discussed quite openly. It has been used by military and civilian organisations so can hardly be classed as a military secret. If you have been 'suspecting' that such a thing existed you could have looked at the science journals decades ago, or perhaps more recently on Wikipedia, and read all about it.

      The clever part of this is that true radar, passive or active, requires a reflective body (e.g. an aircraft fuselage or a vessel on the sea) in order to function, whereas what is being described here is the ability to detect non-reflective bodies i.e. human beings. This depends not upon a direct echo but in detecting the perturbations of radio waves as bodies move through them. As you can see from the 2 links that I provided from earlier stories (2015) this is not a new idea. However, it is the analysis of the perturbations using either or both Doppler effect and AI that has led to very recent advances in the capabilities of such processing. There is also unconfirmed (at least publicly) speculation that the F117 shot down during the Balkan wars was a result of a passive radar detection i.e. the tracking radar detecting an echo from the target from a entirely separate RF source. Even stealth aircraft are only stealthy at a range of frequencies - not all frequencies. This is evident because they are not stealthy to visible light and so we have all seen them, or at least pictures of them.

      The current detection capabilities are relatively short range - measured in metres or at most 10s of metre - using special equipment in relatively quiet RF environments. This does not, however, suggest that the new capabilities are trivial but we are still at the early stage of this research. The capability that you are suggesting that by using cell towers we could all be subjected to everyday tracking is still fanciful, although I have no doubt that there will be some who are currently investigating how quickly they could make such a thing possible. For a start the power of any signal decays quickly dependant on range and, because we are not detecting an echo but changes in RF fields, it is necessary to know how the field behaves when viewing an empty room and comparing that with the changes caused by people moving. It is not as simple as traditional radar where an echo indicates a target and the lack of an echo might indicate that there is no target (but cannot guarantee that there is none - it might just not be reflective at the frequency being used for detection). For what you are suggesting it would require each home, and possible each room in each home, to be 'calibrated' in some way using directional and sensitive sensors close to the physical structures to rule out any perturbations caused by other bodies moving outside the target of interest. For example, birds, dogs, the postman, passers-by etc between the area being observed and the detector will all be having some effect on the RF signals. Directional antennae will help but even then will only help in a very limited fashion over the ranges that you are concerned about.

      This is more likely to be initially of interest to the intelligence community and law enforcement, where they are 'observing' individuals in a small contained area. Being able to see what is on the other side of a wall using entirely passive means (i.e. undetectable by the targets themselves) has several obvious uses. Being able to track everybody in the area covered by, say, a cell tower is a much more difficult nut to crack. You can relax the tightness of your tinfoil hat for the moment.

      • (Score: 2) by corey on Sunday October 13 2019, @09:02PM (1 child)

        by corey (2202) on Sunday October 13 2019, @09:02PM (#906705)

        There is also unconfirmed (at least publicly) speculation that the F117 shot down during the Balkan wars was a result of a passive radar detection i.e. the tracking radar detecting an echo from the target from a entirely separate RF source.

        I think what you're referring to is not passive radar but bistatic radar. It's good as long as your transmitters don't get blown and the target is flying around or over them.

        https://en.m.wikipedia.org/wiki/Bistatic_radar [wikipedia.org]

        Also you mentioned frequencies; low frequency radar will light up a stealth target. But resolution is limited.

        • (Score: 2) by janrinok on Monday October 14 2019, @08:37AM

          by janrinok (52) Subscriber Badge on Monday October 14 2019, @08:37AM (#906872) Journal

          We are both correct.

          A bistatic radar is one where the radar transmitting and receiving antennae are located separately, but they comprise parts of a single radar system. Indeed in some some bistatic systems a single receiver might control several transmitters widely separated to improve radar coverage, avoid interference from other transmitters or to protect against the destruction of a single transmitter. Knowing where the receiver is located is a harder problem to solve and thus less likely to be attacked.

          A passive radar [wikipedia.org] is one where the transmitter is not related to the receiving element i.e. it uses signals that are intended for a different purpose entirely e.g. cell towers, indoor routers, etc. Passive radar is a special case of a bistatic radar.

      • (Score: 2) by jmichaelhudsondotnet on Monday October 14 2019, @04:29PM

        by jmichaelhudsondotnet (8122) on Monday October 14 2019, @04:29PM (#907003) Journal

        I misspoke, I am aware of passive rf radar but the imaging thing is something I have been seeking confirmation for. It is not as if this stuff is easy to get information on either. And I dont claim to be an rf expert but I did pass ee216 so I know a lot more than most about the fourier transform and frequency analysis.

        Thanks though this is very informative and I learn, you clearly know what you are talking about. There is no way that any individual human can keep track of the threats in each field of tech, this is some of the problem we face ever having any rights in the face of this stuff.

        Since in every other case where the government has had the technical capability to steal our rights by simply installin an upgrade, they have done so, which is why I have gotten pretty skeptical of upgrades and new tech, as you have probably deduced by my saying so over and over. The 100 microphones forming an array to get specific very quiet signals is very close to magic, and I dont see why they couldnt do this with 100 cell phone mics in an open area with telemetry and sharp software. Then I wonder if you know the telemetry of 100 wifi access points and 3 towers, and put a few autists on the job, if you could get a back and white of a room that had been calibrated during a covert entry.

        And what if you, idk, had the contract to remodel the white house or 'freedom' tower, could you just build this into the design? Remote imaging without any physical devices laying around is the holy grail of spying, is it not? Not much spies wont do for the holy grail if I recall my indiana jones lore correctly.

        Also nice to hear someone address my actual points rather than simply accusing me of insanity. It is after all why I am here and not other places, even if only 10% of the time there is such a positive result.

        If all I do is give someone the writing prompt to explain something, it is well worth my time. We are doing the work of public interest technologists even if no one else realizes it yet.

      • (Score: 2) by jmichaelhudsondotnet on Monday October 14 2019, @04:34PM

        by jmichaelhudsondotnet (8122) on Monday October 14 2019, @04:34PM (#907007) Journal

        Idk maybe I was writing in a way to sound like I know more than I did, I can do better. Overstating, using bigger scoped words than necessary, I have had trouble with this in the past.

        So easy to do until someone rolls by who knows more.

  • (Score: 0) by Anonymous Coward on Sunday October 13 2019, @02:41PM (1 child)

    by Anonymous Coward on Sunday October 13 2019, @02:41PM (#906625)

    Now police can shoot suspects even if they are hiding behind walls

(1)