Abe Davis: New video technology that reveals an object's hidden properties
آبي دافيز: تقنية الفيديو الجديدة التي تكشف الخصائص المخفية للكائن
Computer vision expert Abe Davis pioneers methods to extract audio from silent digital videos, even footage shot on ordinary consumer cameras. Full bio
Double-click the English transcript below to play the video.
as a very visual thing.
or gesture with my hands while I speak,
that's too subtle for the human eye,
even when humans can't.
of a person's wrist,
of a sleeping infant,
that these were videos,
at two regular images,
almost completely still.
of subtle motion going on here,
the wrist on the left,
the infant on the right,
and fall of her chest
a lot of significance,
too subtle for us to see,
what they call a motion microscope,
these subtle motions in video
become large enough for us to see.
on the left video,
this person's heart rate.
on the right video,
that this infant takes,
to monitor her breathing.
because it takes these phenomena
to experience through touch
and non-invasively.
with the folks that created that software,
that we can use software
as a way to extend our sense of touch.
with our ability to hear?
to capture the vibrations of sound,
into a microphone?
in perspective for you.
work by converting the motion
into an electrical signal,
to move readily with sound
and interpreted as audio.
too subtle and too fast for us to see.
with a high-speed camera
to extract tiny motions
what sounds created them?
into visual microphones from a distance.
that you see on the right
played this sound.
of frames per second,
just sitting there doing nothing,
by about a micrometer.
a hundredth and a thousandth
perceptually invisible.
can be perceptually invisible
seemingly still video
out of so little motion?
move by just a single micrometer,
by just a thousandth of a pixel.
of pixels in it,
of the tiny motions that we see
to something pretty significant.
when we figured this out.
a pretty important piece of the puzzle.
that affect when and how well
and the lens that you use;
and how loud your sound is.
with our early experiments,
any of these factors wrong,
what the problem was.
experiments looked like this.
see our high-speed camera,
by these bright lamps.
very careful in these early experiments,
Little lamb! Little lamb!
looks completely ridiculous.
we tried this on. (Laughter)
to recover this sound.
Little lamb! Little lamb!
we recovered intelligible human speech
to modify the experiment,
or moving the object further away,
the limits of our technique,
to a bag of chips,
about 15 feet away,
by only natural sunlight.
from inside, next to the bag of chips.
whose fleece was white as snow,
that lamb was sure to go.
to recover from our silent video
whose fleece was white as snow,
that lamb was sure to go.
that we can push these limits as well.
plugged into a laptop computer,
the music that was playing on that laptop
by changing the hardware that we use.
I've shown you so far
a high-speed camera,
about a 100 times faster
to use this technique
of what's called a rolling shutter.
record images one row at a time,
during the recording of a single image,
between each row,
is that by analyzing these artifacts,
using a modified version of our algorithm.
music from before,
store-bought camera,
the sound that we recovered,
distorted this time,
recognize the music.
is that we were able to do this
that you could literally run out
about surveillance.
this technology to spy on someone.
a lot of very mature technology
from a distance for decades.
to picture the vibrations of an object,
through which to look at the world,
that cause an object to vibrate,
the ways that we use video,
to look at things,
that we learn about the world:
still won't let us do,
just a few months ago,
I've shown it to a public audience.
to use the vibrations in a video
that will let us interact with them
in the shape of a human,
with just a regular camera.
about this camera.
with my cell phone before.
on the surface where it's resting
of regular video,
the vibrations in that video
and material properties of our object,
to create something new and interactive.
and it's not a video,
with the object.
that we've never seen before,
five seconds of regular video.
way to look at the world,
how objects will respond
looking at an old bridge
how would that bridge hold up
that you probably want to answer
across that bridge.
limitations to this technique,
with the visual microphone,
in a lot of situations
here's a video that I captured
to create this simulation.
to a film director,
in a shot after it's been recorded.
at a hanging curtain,
any motion in this video,
imperceptible motions and vibrations
to create this simulation.
this kind of interactivity
and 3D models,
from real objects in the real world
a lot of potential.
who worked with me on these projects.
is only the beginning.
with this kind of imaging,
with common, accessible technology.
really exciting to explore
ABOUT THE SPEAKER
Abe Davis - Computer scientistComputer vision expert Abe Davis pioneers methods to extract audio from silent digital videos, even footage shot on ordinary consumer cameras.
Why you should listen
MIT PhD student, computer vision wizard and rap artist Abe Davis has co-created the world’s most improbable audio instrument. In 2014, Davis and his collaborators debuted the “visual microphone,” an algorithm that samples the sympathetic vibrations of ordinary objects (such as a potato chip bag) from ordinary high-speed video footage and transduces them into intelligible audio tracks.
Davis is also the author of Caperture, a 3D-imaging app designed to create and share 3D images on any compatible smartphone.
Abe Davis | Speaker | TED.com