Supasorn Suwajanakorn: Fake videos of real people -- and how to spot them
Supasorn Suvajanakorn (Supasorn Suwajanakorn): Lažni snimci pravih ljudi - i kako ih uočiti
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.
porodicama s refinansiranjem domova,
refinance their homes,
like high-tech manufacturing,
visoko-tehnološka proizvodnja,
koja stvara dobre nove poslove.
that creates good new jobs.
kako smo dospeli ovde.
za ovaj posao bio je projekat
chance for learning about the Holocaust
za saznanje o holokaustu od preživelih.
interactive conversations
interaktivne razgovore
of a real Holocaust survivor.
preživelog iz holokausta.
survive the Holocaust?
preživeli holokaust?
were prerecorded in a studio.
unapred snimljeni u studiju.
and to him as a person.
njegovoj priči i njemu kao osobi.
about human interaction
u vezi sa ljudskom interakcijom
or movies could ever teach us.
ikada mogli da nas nauče.
like this for anyone?
ovakav model za bilo koga?
and acts just like them?
i ponaša se baš kao oni?
da li je izvodljivo
using nothing but these:
koristeći ništa drugo osim ovoga:
this kind of passive information,
ovu vrstu pasivnih informacija,
a Nobel Prize winner in physics
Nobelovu nagradu za fiziku,
kao legendaran predavač.
if we could bring him back
kada bismo mogli da ga vratimo
and inspire millions of kids,
i inspiriše milione klinaca,
but in any language?
već na bilo kom jeziku?
for advice and hear those comforting words
i deke za savet i čujete utešne reči
book authors, alive or not,
pisci knjiga, živi ili ne,
for anyone interested.
sve svoje knjige za sve zainteresovane.
here are endless,
3D face model from any image
veoma detaljan 3D model lica
from different views.
iz različitih uglova.
on each video frame
na svakoj slici snimka
output model from different angles.
iz različitih uglova.
is very challenging,
veoma problematično,
is that we are going to analyze
veliku zbirku fotografija osobe.
of the person beforehand.
we can just search on Google,
da potražimo na Guglu
to build an average model,
da napravimo prosečan model,
to recover the expression
kao što su brazde i bore.
like creases and wrinkles.
can come from your typical photos.
od vaših tipičnih fotografija.
what expression you're making
that there are a lot of them.
a new blending technique
a single averaging method
metodu uprosečavanja
facial textures and colors.
of a model of a person,
is by a sequence of static photos.
je preko niza nepokretnih fotografija.
depending on the expression.
kako se bore pojavljuju i nestaju
to drive the model.
da upravljamo modelom.
ali nekako smo uspeli
some more amazing people.
are controllable models
from their internet photos.
iz njihovih internet fotografija.
the motion from the input video,
pokret sa ulaznog snimka
da upravljamo celom družinom.
„Ovaj nacrt zakona će teško proći
It's a difficult bill to pass,
is to capture their mannerisms
da uhvatimo njihovu mimiku
of these people talks and smiles.
od ovih ljudi priča ili se smeši.
actually teach the computer
možemo li zaista naučiti računar
video footage of the person?
I let a computer watch
giving addresses.
i njegovih obraćanja.
given only his audio.
samo iz njegovog zvučnog snimka.
14.5 million new jobs
14,5 miliona novih poslova
is only the mouth region,
je samo oblast usta,
into these mouth points.
u ove tačke usta.
or through Medicare or Medicaid.
ili preko zdravstvene zaštite.
enhance details and teeth,
poboljšavamo detalje i zube
and background from a source video.
iz izvornog snimka.
besplatno da se pregledaju
just for being a woman.
samo zato što ste žena.
on a parent's plan until they turn 26.
u roditeljskom programu do 26. godine.
seem very realistic and intriguing,
izgledaju vrlo realistično i interesantno,
frightening, even to me.
čak i za mene.
of a person, not to misrepresent them.
precizan model osobe,
is its potential for misuse.
je potencijal za zloupotrebu.
about this problem for a long time,
first hit the market.
on countermeasure technology,
na tehnologiji za protivmeru
effort at AI Foundation,
u fondaciji za veštačku inteligenciju
of machine learning and human moderators
mašinskog učenja i ljudskih moderatora
is called Reality Defender,
se zove „Branilac stvarnosti“ -
that can flag potentially fake content
da označi potencijalno lažne sadržaje
da učine dosta štete
stigne da ih proveri,
that we make everyone aware
and be critical about what we see.
i da posmatramo kritički.
we can fully model individual people
da možemo potpuno da modelujemo pojedince
the safety of this technology.
bezbednost ove tehnologije.
na pravi način i pažljivo,
uticaj ma kog pojedinca na svet
positive impact on the world
the way we want it to be.
budućnost u onakvu kakvu želimo.
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