Follow Slashdot stories on Twitter

 



Forgot your password?
typodupeerror
×
Media Piracy Technology Your Rights Online

Mpeg 7 To Include Per-Frame Content Identification 273

An anonymous reader writes "NEC has announced that its video content identification technology has been incorporated in the upcoming Mpeg 7 video standard, allowing for each video frame to have its own signature, meaning that even minute changes to the file such as adding subtitles, watermarks or dogtags, and of course cutting out adverts, will alter the overall signature of the video. According to NEC this will allow the owners of the video to automatically 'detect illegal copies' and 'prevent illegal upload of video content' without their consent. NEC also claims that its technology will do away with the current manual checking by members of the movie industry and ISPs to spot dodgy videos."
This discussion has been archived. No new comments can be posted.

Mpeg 7 To Include Per-Frame Content Identification

Comments Filter:
  • Re:modest proposal (Score:3, Interesting)

    by andi75 ( 84413 ) on Monday May 10, 2010 @02:08AM (#32152122) Homepage

    I would mandate the opposite legislation. Any device that's sold or rented to consumers must also include all contained cryptographic keys in an easily accessible manner (e.g. on an accompagning CD). That way it is guaranteed that consumers can always, and without limitations, accesss the data they paid for.

  • by Anonymous Coward on Monday May 10, 2010 @02:08AM (#32152124)

    So it sounds like the easy way to upload "protected" content would be a quick transcode with a slightly different bitrate, thereby removing the per frame signature, causing it to be unrecognizable by the automated checker...

  • Re:modest proposal (Score:2, Interesting)

    by mb1 ( 966747 ) on Monday May 10, 2010 @02:13AM (#32152154)

    I'm sure looking forward to the future when I'll be prosecuted by the patent and/or trademark holders of both 'Do Nothing' and 'Do Something' for doing something and/or doing nothing at all.

    Of course, Apple will offer their 'Doing Apple' to give us all some choice - but at the same time will sue anyone trying to 'Do Nothing' or 'Do Something' and not 'Doing Apple'...

  • I can't wait (Score:4, Interesting)

    by masterwit ( 1800118 ) * on Monday May 10, 2010 @02:14AM (#32152166) Journal

    A new algorithm to crack, Math is Fun! (They don't realize that some of us do this as a passion, no I endorse fully supporting those companies that deserve it, but not everyone does this for piracy, its just a hella lotta fun to crack the reported "uncrackable".)

    Just my take, I love math.

  • Backwards or not? (Score:1, Interesting)

    by Anonymous Coward on Monday May 10, 2010 @02:22AM (#32152222)

    On the first sight, it looks like the stupidest idea ever. You can't use digital signatures to protect from the current considered dangers (piracy, re-use, ripping). People will happily remove all signatures and edit the media like the wish.

    I think the real issue is what they are going to use MPEG 7 _with_. Expect heavy DRM and content access restrictions. _Then_ the signatures will play a vital role; you will not be able to play anything that was not signed.

  • by Anonymous Coward on Monday May 10, 2010 @02:33AM (#32152296)

    It's funny how they assume more signatures will change anything, the data stored inside a video format, no matter how restrictive and closed, will eventually be converted to a less restrictive format, stripped of all the unwanted stuff, this only adds yet another inceptive to do so, realtime packet inspection (how else would ISPs check signatures of video frames?) to determine what files are transmitted in realtime sounds nice on paper, until you factor in 8192 bit encryption and the fact you can make a video look like any other bit of random binary garbage data rather easily (I know, lets outlaw any files not whitelisted by the MPAA!!). The only thing this would effect (as usual) are people who obtain the video legally and something (minor disk write error, scratch on another medium etc) alters a single bit in the file, thus making the entire thing appear invalid to any player or system that would enforce this implementation.

  • by fuzzyfuzzyfungus ( 1223518 ) on Monday May 10, 2010 @02:39AM (#32152342) Journal
    The secret sauce proprietary algorithm in the (puff piece) TFA sounds like a file verification mechanism, in the vein of CRC, hash verification, and friends. Which is odd; because the problem of keeping digital data reasonably uncorrupt is a serious one for Big Storage type outfits, and archivists; but it hasn't been much of a concern for team content. What they've wanted is watermarks, "traitor tracing", and all that. Now, a good verification algorithm is a terrible watermark algorithm, and vice versa, period. Verification algorithms are supposed to freak out if so much as a single bit has been twiddled. Watermark algorithms are supposed to be robust against common forms of tampering and re-encoding.

    So, what's the deal?

    1. It could be that "PC Authority" has been handed an NEC press release, and can't even handle the challenge of regurgitating it properly. In which case, any speculation based on the details of TFA is pointless, if TFA is so much commercial word salad.

    2. It could also be that PC Authority is reading the NEC release more or less correctly; but the release was just blitzed out by some PR flack, and they lack the context. This is, in fact, an integrity verification technology, designed to work quickly on video streams, that will be included in some future standard, as an obscure convenience to future editors and producers and archivists who will have to deal with 10,000 hours of MPEG7 video in OMG-4k-Super-def-3D, and need to know, fast, if any of it is getting munged. It would be a super boring, highly specific part of the spec, of basically no interest to the general public; but it could be more or less true as described.

    3. And here's the sinister conspiracy theory: Where do file integrity verification and DRM come together? If, and only if, planned devices are "default deny, play signed content only". If your Blu-ray2 player simply refused to play anything that isn't a wholly unaltered copy of a commercial release, the otherwise absurd(as noted above) notion that an integrity check algorithm can serve as a piracy deterrent becomes true... It wouldn't stop cammer kiddies from playing altered copies on general purpose PCs, if those are still alive; but making "blessed only" a condition of the licencing agreement for future STB-type devices would basically kill the unsophisticated pirated disk market(barring hardware hacks on specific devices, or really stupid mistakes in media design).
  • by Protoslo ( 752870 ) on Monday May 10, 2010 @03:02AM (#32152426)

    The firm touts the efficiency of its algorithm, saying that a bog standard PC can search through 1,000 hours video in just one second. Quite what the firm's definition of a "home-class" PC would be interesting to know as we can't quite figure out how even a dual core 3GHz box can go through the 104 billion checks for 1,000 hours of video in a mere second.

    1000 hours of video has close to 104 million frames; that would yield around 60 cycles per frame on a dual core (i.e. old) box.

    The innumeracy of the author aside, what does this technology even do? Apparently altering the video, even minutely, will alter the "signature." Much like...CRC-32...very cutting-edge. We should name this startling development; I nominate the word "hash." Stupefied by the summary and the "article," I turned to the actual press release [nec.co.jp] to find out what the technology really (purportedly) does.

    1. Accurate detection of copied or altered video content Video signatures are extracted for each frame based on differences in the luminance between sets of sub-regions on a frame that are defined by a variety of locations, sizes, and shapes. Video signatures represent a unique fingerprint that can be individually detected frame by frame. This technology is capable of accurately detecting video content with that was created with such editing operations as analog capturing (*3), re-encoding (*4) and caption overlay (*5), which was conventionally very difficult to detect.
    ...
    4. Compatibility with home PCs By designing a compact signature size of 76 bytes per frame, the storage memory required for the matching process is minimized. As a result, a home-class PC (*8) can match approximately 1,000 hours of video in 1 second.

    It turns out that a home-class PC ("A single core CPU with 3GHz clock speed was used for testing purposes. Signatures were stored in the main memory.") is able to match 1000 hours that have already been hashed in a single second. No doubt it takes considerably longer to actually calculate the signatures. The power of the algorithm is that when the video is altered (in human-recognizable ways) the signature doesn't change much. Ah, things are starting to actually make sense. The truth is (surprise!) the opposite of the linked phrase in the summary.

    This technology may allow automated, accurate matching of copyrighted video on youtube or other video sites...who cares? That is already being done, only less accurately. The law would have to change rather drastically for it to be mandated that everyone includes correct hashes in their MPEG-7 video. That is hardly necessary--I'm sure someone will spare the cycles to hash the videos and inform content owners. Like they do now...only better. Maybe next time we can all have fun panicking about the "FaceRecognition descriptor" [webstore.iec.ch] (only the TOC/summary is free) instead. Really, the 76-byte signature is just an implementation of the metadata schema for MPEG-7. The algorithm should work for any format, however (otherwise it would be rather trivial to evade!).

    The only interesting thing I have learned is that NEC's algorithm uses robust, compactly representable edge detection (maybe) to compare short clips of video with extremely high accuracy; yay, computer science. All of this escaped Lawrence Latif, author of TFA (such as it is), who didn't see fit to RTFA himself before he started blogging his paranoid fantasies as fact. I wonder just who the "anonymous reader" that submitted the summary was?

  • by purpledinoz ( 573045 ) on Monday May 10, 2010 @04:13AM (#32152638)
    The invisible hand of market forces is slapping the shit out of these companies. I wonder when they're going to realize that it's pointless fighting against it. Adding features that takes away value is no way to win customers. I would LOVE to PAY for an online streaming service, where I have access to all TV shows and movies with any choice of subtitles, and dubbed languages available. So far, since I'm in Germany, I found MaxDome, where I can only watch a limited selection of movies dubbed only in German, no subtitle options. What if I want to watch a movie that's not very well-known? I can either order it from Amazon, or just Google the movie title and stream it.
  • by mrrudge ( 1120279 ) on Monday May 10, 2010 @04:39AM (#32152758) Homepage
    Credit card numbers are a bad example as they're a piece of information which is generally transferred securely between two parties who have motive for them to remain private. If credit card numbers were a product which was distributed to many people even one of whom fails to keep them private you would be able to access them on node B.

    Furthermore, if node B has internet access and it's user be sufficiently lacking in morals and know where to look, it's entirely possible that credit card numbers could be found from there ( as far as I understand, buyable in large batches ) and it would currently not be possible for the network to recognise and stop the movement of these bits.
  • Re:modest proposal (Score:4, Interesting)

    by JWW ( 79176 ) on Monday May 10, 2010 @07:42AM (#32153462)

    Actually if we could convince the federal government to be as lax in enforcing IP laws as they are with enforcing immigration laws, we'd be in great shape.

    Its interesting that immigration laws are very very poorly enforced and yet the gov't doesn't really want to fix the situation, but the fact that someone might download a movie and watch it, omg release the hounds!!!

    Everyone in Arizona is getting all up in arms about a law asking evernone to present proof of citizenship papers. In this IP battle, they're consistently asking us to provide proof of purchase papers at every step to view content, and no one in the general public bats an eye.

    But there is one obvious parallel here. In both immigration and IP law the government we're getting is acting exactly how the big companies in control want it to.

  • owners (Score:4, Interesting)

    by jDeepbeep ( 913892 ) on Monday May 10, 2010 @09:42AM (#32154506)

    According to NEC this will allow the owners of the video to automatically 'detect illegal copies' and 'prevent illegal upload of video content' without their consent.

    If I bought it, that owner is me.

  • Re:modest proposal (Score:3, Interesting)

    by Jackie_Chan_Fan ( 730745 ) on Monday May 10, 2010 @02:09PM (#32159200)

    Its interesting because I think we have made demands. They just are not answered.

    So we continue to download mkv rips of bluray for example.

    We're just not paying attention to their offerings anymore because they have not met the demand. We've made the demands, they didnt listen.. and now we dont care.

    I do think the industry has been listening somewhat. The free DVD version of movies that come with new Bluray Releases is a nice way of listening to the public's cry about buying 1 format, and having to buy another. At least now they give you a DVD version when you buy the bluray... well at least only a few of the newer titles.

    Anyways.. The people have spoken and in order for them to find freedom, they had to become criminals.

Intel CPUs are not defective, they just act that way. -- Henry Spencer

Working...