• Rouslan Galioullin

    future is atomic particles

  • http://www.facebook.com/people/Ryan-Bridgeman/699795034 Ryan Bridgeman

    I was thinking about something related to this article the other day (although I just saw it just now).

    Of all of the video intelligence in the world, either fed to the government or fed within a closed environment, there is a surplus.

    I don’t really know how long data is stored for a convenient store, shopping mall, etc. but NVIDIA GPU’s could be used to identify “people in the wrong” by scanning video and looking for certain faces using typical face recognition.  The amount of video data out there would take months or even years to sift through by humans to find connections to human subjects but I believe CUDA could make a big difference in this respect.  The only problem is the quality of video in these organizations.  Obviously with quality comes filesize.

    Imagine being part of the police / government and wanting to identify a person and the businesses/locations they’ve been to in the last 24 hours.  This could take awhile.  However, with video parallel processing, it could take up to a few hours.  I am unaware something like this exists but it would definitely speed up the process.

  • Phan Thanh

    GE Intelligent Platforms is implementing a family of rugged products based around the NVIDIA® CUDA™ architecture. With a vast resource of reference material and wide range of proven application use, CUDA is set to revolutionize the computing capability within the Mil/Aero application space.

  • Will Park

    Interesting point, you make there, Ryan. Sounds like GPUs might be able to speed up the process you mention.

  • http://www.nvidia.com Michael Steele

    Great point. There are already several video applications using CUDA to do exactly what you suggest… detect and recognize specific objects or faces, track movement of people or things (called “Change Detection”), and even enhance the video “CSI-style” by improving resolution, stability, sharpening, etc…  Saving time is the key here, whether your goal is to view and analyze in real-time or simply shorten the time it takes to process hours of video. You’ll also find a growing list academic research activities in this area, so I’m sure you’ll see a lot more of these groundbreaking video tools make their way to NVIDIA GPUs.

  • http://www.nvidia.com Michael Steele

    The GE platforms, specifically the MAGIC1, are a great way to get high performance into the hands of those in the field, where the speed of response means everything.  Saving time often means saving lives.  It’s also a great example of the “green” value of GPU computing.  Parallel processing on this scale lets you get a ton of processing capability into a small space without requiring excess power.