Supasorn Suwajanakorn: Fake videos of real people -- and how to spot them
Supasorn Suwajanakorn: Tikrų žmonių netikri vaizdo įrašai, ir kaip juos atpažinti
Supasorn Suwajanakorn works on ways to reconstruct, preserve and reanimate anyone -- just from their existing photos and videos. Full bio
Double-click the English transcript below to play the video.
kuris iš Obamų yra tikras:
refinance their homes,
refinansuoti būsto paskolas,
like high-tech manufacturing,
that creates good new jobs.
naujų gerų darbo vietų.“
netikras?
nėra tikras.
projektas,
chance for learning about the Holocaust
sužinoti apie Holokaustą
liudijimų dimensijos“ .
interactive conversations
galimybę,
of a real Holocaust survivor.
Holokaustą, holograma.
survive the Holocaust?
Holokaustą?“
were prerecorded in a studio.
įrašyti studijoje,
and to him as a person.
ir su juo, kaip asmenybe.
about human interaction
padaro žmonių bendravimą
or movies could ever teach us.
taip nesudomintų.
like this for anyone?
asmeniui?
and acts just like them?
taip, kaip tas žmogus?
using nothing but these:
modelį vien tik iš
bei vaizdo įrašų.
this kind of passive information,
pasyvią informaciją t. y.
asmenį.
a Nobel Prize winner in physics
laureatas fizikos srityje,
if we could bring him back
galėtume jį susigrąžinti
and inspire millions of kids,
milijonus vaikų?
but in any language?
anglų kalba?
for advice and hear those comforting words
patarimo savo mirusių senelių ir
book authors, alive or not,
išgirsti gyvus ir jau mirusius
for anyone interested.
here are endless,
3D face model from any image
iš bet kokios nuotraukos
from different views.
iš skirtingų pusių.
on each video frame
pats algoritmas
output model from different angles.
is very challenging,
is that we are going to analyze
of the person beforehand.
prieš pradedant kurti modelį.
we can just search on Google,
to build an average model,
to recover the expression
kuris atkurtų veido išraišką,
like creases and wrinkles.
nelygumai ir raukšlės.
can come from your typical photos.
jūsų įprastos fotografijos,
what expression you're making
that there are a lot of them.
tokių nuotraukų.
a new blending technique
a single averaging method
nei suvidurkinimo metodas,
facial textures and colors.
pavšių ir spalvas.
išraiška.
of a model of a person,
galime valdyti
is by a sequence of static photos.
depending on the expression.
tai atsiranda, tai išnyksta.
to drive the model.
naudoti vaizdo įrašą.
kažkaip
some more amazing people.
nuostabių žmonių.“
are controllable models
nuotraukų
from their internet photos.
the motion from the input video,
vaizdo įrašo,
It's a difficult bill to pass,
sunku,
nemalonūs.“
is to capture their mannerisms
įamžinti žmonių manierizmus
of these people talks and smiles.
kalbant ar šypsantis.
actually teach the computer
video footage of the person?
I let a computer watch
peržiūrėti
giving addresses.
sako savo kalbas.
given only his audio.
balso įrašą.
14.5 million new jobs
14,5 mln. darbų
is only the mouth region,
into these mouth points.
ir pritaikymui burnoms.
or through Medicare or Medicaid.
arba iš sveikatos draudimo.
enhance details and teeth,
patobuliname detales, dantis
and background from a source video.
fonu iš pirminio video.
pasitikrinti sveikatą
just for being a woman.
įtakos kainai.
on a parent's plan until they turn 26.
iki 26 metų.
seem very realistic and intriguing,
ir intriguojantys,
frightening, even to me.
of a person, not to misrepresent them.
modelį, o ne jo kopiją.
is its potential for misuse.
tai bus netinkamai panaudota.
about this problem for a long time,
first hit the market.
on countermeasure technology,
grėsmių neutralizavimo technologiją.
effort at AI Foundation,
of machine learning and human moderators
intelektą
is called Reality Defender,
„Realybės gynėjas“.
that can flag potentially fake content
galintis pranešti apie suklastotą
tiesiogiai naršyklėje.
daug žalos
that we make everyone aware
apie tai,
and be critical about what we see.
tai, ką matome.
we can fully model individual people
bei patikimai šios technologijos apsaugai
the safety of this technology.
šio įrankio naudojimas
positive impact on the world
poveikį
the way we want it to be.
kaip norime.
ABOUT THE SPEAKER
Supasorn Suwajanakorn - Computer scientistSupasorn Suwajanakorn works on ways to reconstruct, preserve and reanimate anyone -- just from their existing photos and videos.
Why you should listen
Can we create a digital avatar that looks, acts and talks just like our sweet grandma? This question has inspired Supasorn Suwajanakorn, a recent PhD graduate from the University of Washington, to spend years developing new tools to make it a reality. He has developed a set of algorithms that can build a moving 3D face model of anyone from just photos, which was awarded the Innovation of the Year in 2016. He then introduced the first system that can replicate a person's speech and produce a realistic CG-animation by only analyzing their existing video footage -- all without ever bringing in the person to a Hollywood capture studio.
Suwajanakorn is working in the field of machine learning and computer vision. His goal is to bring vision algorithms out of the lab and make them work in the wild.
Supasorn Suwajanakorn | Speaker | TED.com