Stories
Slash Boxes
Comments

SoylentNews is people

posted by mrpg on Friday February 10 2017, @06:48AM   Printer-friendly
from the I-saw-it-coming dept.

Google has developed a neural network algorithm that can create reasonable approximations of a 32×32 image from a downsized 8×8 image:

Of course, as we all know, it's impossible to create more detail than there is in the source image—so how does Google Brain do it? With a clever combination of two neural networks. The first part, the conditioning network, tries to map the the 8×8 source image against other high resolution images. It downsizes other high-res images to 8×8 and tries to make a match.

The second part, the prior network, uses an implementation of PixelCNN to try and add realistic high-resolution details to the 8×8 source image. Basically, the prior network ingests a large number of high-res real images—of celebrities and bedrooms in this case. Then, when the source image is upscaled, it tries to add new pixels that match what it "knows" about that class of image. For example, if there's a brown pixel towards the top of the image, the prior network might identify that as an eyebrow: so, when the image is scaled up, it might fill in the gaps with an eyebrow-shaped collection of brown pixels. To create the final super-resolution image, the outputs from the two neural networks are mashed together. The end result usually contains the plausible addition of new details.

Pixel Recursive Super Resolution

There is an excellent write-up on Wikipedia about various techniques for creating face hallucinations — aka an enhanced-resolution image of a face.


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This discussion has been archived.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.
  • (Score: 2) by takyon on Friday February 10 2017, @05:18PM

    by takyon (881) <takyonNO@SPAMsoylentnews.org> on Friday February 10 2017, @05:18PM (#465535) Journal

    Given how much details can be extracted out of the 8x8 image, I think that algorithms can do a lot better if they are given more, like 64x64.

    Anyway, what is more likely to happen than your scenario of the defense arguing over the algorithm is something like parallel construction. You get picked up by the mass facial recognition network or mass surveillance network. The police/feds interrogate you until you admit to the crime or accidentally give them more evidence. You go to court, and you accept a plea bargain. Case closed. No need to attack the algorithm... it was just a way to get you into custody. If they can hoover up all the images published to social media in real time, and combine it with other stuff like Stingray, CCTV, and license plate scanners, they can even catch the most wanted suspects living in the middle of nowhere who have the need to stop at a WalMart once in a while.

    --
    [SIG] 10/28/2017: Soylent Upgrade v14 [soylentnews.org]
    Starting Score:    1  point
    Karma-Bonus Modifier   +1  

    Total Score:   2  
  • (Score: 0) by Anonymous Coward on Friday February 10 2017, @05:36PM

    by Anonymous Coward on Friday February 10 2017, @05:36PM (#465540)
    I know what you are saying, but the law has a clear case for this: The Police Sketch. It is a "best guess" of what someone looks like, is enough to try to find the person that is being looked for, is enough to start an investigation, and is NOT enough to convict. "You look like this police sketch, but camera, GPS, and testimony puts you out to dinner with your wife" is still reasonable doubt.
  • (Score: 3, Informative) by DannyB on Friday February 10 2017, @07:44PM

    by DannyB (5839) Subscriber Badge on Friday February 10 2017, @07:44PM (#465570) Journal

    Parallel Construction: Conspiracy of prosecutors and law enforcement to commit perjury by lying to the court and the defense about what their evidence actually is.

    --
    To transfer files: right-click on file, pick Copy. Unplug mouse, plug mouse into other computer. Right-click, paste.