Predator Camera Learns, Shows What Future Autofocus Might Be Like

Computer vision PhD student Zdenek Kalal developed a camera called “Predator” that learns from its mistakes while given difficult recognition and tracking tasks. It’s a pretty interesting glimpse at how powerful the autofocus feature on consumer cameras might be in the future. Imagine being able to teach your camera to recognize a particular moving subject, then having all the photographs taken show that subject in perfect focus! Cameras already have simple facial recognition features built in these days, but something like this would take it to the next level.

Head on over to Kalal’s project page, where you can even download the software to try out yourself.

Tracking-Learning-Detection (TLD) (via Wired)

  • mark

    there are so many military aspects to this type of software its just amazing…

  • Brett

    Would be nice to use this in a live environment for a video shoot so you can get a quick draft of some tracking effects.

  • Chung Dha Lam

    Its not that new I already have webcam software which can track my head and put a mask over my face. All the stuff he is saying is already done. Only interesting was tracking to stabilize in real time, but I think mostly looks like same basic of motion tracking in after effects.

  • Pingback: Predator: A Smart Camera that Learns | the fabulous freak()

  • Peter

    Your webcam software won’t be able to track your face if you turn your head sideways or if your face disappears from the screen. This one can, and I think it accomplishes the feat by “learning” or saving all the different perspectives of the object frame by frame as it moves realtime. Also, if you hadn’t noticed, the software was also able to recognize the guy in the picture as well.

  • Fms3691

    This is nothing new. There are at least two U.S. companies that have been using this technology amongst other tools for industrial image processing for years (at least a decade), i.e. for applications in the industrial automation environment.

  • Bastion

    Yeah, but this is Open Source.