GPUs Can Track You in 4K

Discussion in '[H]ard|OCP Front Page News' started by AlphaAtlas, Nov 8, 2018.

  1. AlphaAtlas

    AlphaAtlas Gawd Staff Member

    Messages:
    1,015
    Joined:
    Mar 3, 2018
    GPUs are already being deployed to track people outside of Teslas and on the streets, but apparently, the algorithms those GPUs run are limited to relatively low resolutions of about 608x608 pixels. To keep up with improving camera technology, researchers have come up with a new algorithm that can handle higher resolution video. "YOLO v2" downscales the large video frames first, and "performs a fast object detection" that it uses to intelligently crop the full resolution frame and send the pieces to multiple GPUs. This yields much better results than just downscaling a feed before processing, and is faster than processing the full frame.

    Check out a video of the algorithm in action here.

    "Our method reduced the time necessary to process high-resolution images by approximately 20 percent, compared to processing every part of the original image under high resolution," Růžička said. "The practical implication of this is that near real-time 4K video processing is feasible. Our method also requires a lower number of server workers to complete this task."
     
  2. theplaidfad

    theplaidfad Lurker

    Messages:
    1,036
    Joined:
    Apr 24, 2008
    The eyes of skynet are getting upgrades.
     
  3. The Mad Atheist

    The Mad Atheist Gawd

    Messages:
    639
    Joined:
    Mar 9, 2018
    "YOLO v2"
    You're Only Living Oppressed
    China gonna luv this!
     
  4. Joust

    Joust [H]ard|Gawd

    Messages:
    1,498
    Joined:
    Nov 30, 2017
    They Live
     
    scojer likes this.
  5. pek

    pek prairie dog

    Messages:
    726
    Joined:
    Nov 7, 2005
    I predict a major research effort into masking your face without actually using a (visible) mask.
     
  6. There is absolutely nothing revolutionary here.

    It's the same "Fuzz" input. See what the angle of difference is to the trained vector to develop a normalized value between 0 and 1 (1 being 100% match) for potential tracking targets. Determine inter-frame velocity and direction based on traied patterns. (Similar to mpeg encoding) Refine if close, and reanalyze on a higher detail level. This is a big old DUH.
     
  7. The Mad Atheist

    The Mad Atheist Gawd

    Messages:
    639
    Joined:
    Mar 9, 2018
    Reminded me of Phantom 2040.
     
  8. Wrecked Em

    Wrecked Em [H]ere for the GangBang

    Messages:
    5,176
    Joined:
    Sep 14, 2004
    24903626.gif
     
    scojer and Aioeyu like this.
  9. SvenBent

    SvenBent 2[H]4U

    Messages:
    2,384
    Joined:
    Sep 13, 2008
  10. Oldmodder

    Oldmodder Limp Gawd

    Messages:
    294
    Joined:
    Aug 24, 2018
    Targets detected, charging phased plasma rifle, selecting full auto.
     
  11. babochee

    babochee Gawd

    Messages:
    584
    Joined:
    Aug 5, 2008
    Kinda struggles with the people rolling down the hill...
     
Tags: