Supasorn Suwajanakorn: Fake videos of real people -- and how to spot them
Supasorn Suwajanakorn: Vídeos falsos de personas verdaderas, y cómo detectarlos
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.
cuál de estos es el Obama verdadero.
a las familias a refinanciar sus casas,
refinance their homes,
like high-tech manufacturing,
that creates good new jobs.
que genera puestos de trabajo.
me inspiré en un proyecto
chance for learning about the Holocaust
oportunidad de saber del Holocausto
dimensiones en testimonio",
interactive conversations
una conversación interactiva
of a real Holocaust survivor.
sobreviviente del Holocausto.
sobrevivió al Holocausto?
survive the Holocaust?
la Providencia velaba por mí.
were prerecorded in a studio.
previamente grabadas en un estudio.
and to him as a person.
con su historia y con él como persona.
about human interaction
humana tiene algo especial.
or movies could ever teach us.
de libros, clases o películas.
como este con cualquier persona,
like this for anyone?
and acts just like them?
y actúe como la persona real?
using nothing but these:
utilizando solo esto que vemos aquí:
this kind of passive information,
información que ya existe,
que encontramos por allí,
modelar cualquier persona.
a Nobel Prize winner in physics
el Premio Nobel de física,
if we could bring him back
and inspire millions of kids,
e inspirara a millones de jóvenes,
but in any language?
for advice and hear those comforting words
y oír esas palabras reconfortantes
book authors, alive or not,
los escritores, vivos o no,
for anyone interested.
a quienes estén interesados.
here are endless,
sumamente interesante.
para reconstruir un rostro en 3D
3D face model from any image
a partir de cualquier imagen,
un escaneo en 3D de la persona.
from different views.
visto desde distintos ángulos.
on each video frame
en cada cuadro del vídeo
output model from different angles.
obtenido, desde distintos ángulos.
is very challenging,
numerosos desafíos,
is that we are going to analyze
en analizar de antemano
of the person beforehand.
we can just search on Google,
basta con buscar en Google.
un modelo de base,
to build an average model,
to recover the expression
para recuperar la expresión
like creases and wrinkles.
can come from your typical photos.
comunes de la persona,
what expression you're making
o el lugar donde se tomó la foto.
that there are a lot of them.
a new blending technique
una nueva técnica de mezcla
a single averaging method
y colores faciales definidos.
facial textures and colors.
a cualquier expresión.
of a model of a person,
el modelo de una persona,
is by a sequence of static photos.
de fotos estáticas.
depending on the expression.
las líneas del rostro según la expresión.
to drive the model.
para manejar el modelo.
pero de alguna manera,
some more amazing people.
are controllable models
from their internet photos.
disponibles en la web.
the motion from the input video,
del vídeo original,
It's a difficult bill to pass,
una ley fácil de aprobar
a veces son difíciles.
is to capture their mannerisms
es captar los gestos típicos
of these people talks and smiles.
y sonreír de cada persona.
actually teach the computer
enseñarle a la computadora
video footage of the person?
fue mostrarle a la computadora
I let a computer watch
giving addresses.
con discursos de Barack Obama.
given only his audio.
tan solo con el audio.
14.5 million new jobs
14,5 millones de puestos de trabajo
is only the mouth region,
solo la parte de la boca.
usa una red neuronal
into these mouth points.
en estos puntos de la boca.
or through Medicare or Medicaid.
nuestro trabajo, Medicare o Medicaid.
enhance details and teeth,
mejoramos los detalles y los dientes,
and background from a source video.
y al fondo del vídeo original.
derecho al chequeo gratuito.
just for being a woman.
solo por ser mujer.
on a parent's plan until they turn 26.
del plan de sus padres hasta los 26.
seem very realistic and intriguing,
son muy realistas y fascinantes,
frightening, even to me.
aterradores, incluso para mí.
of a person, not to misrepresent them.
de una persona, no tergiversarla.
is its potential for misuse.
su potencial de ser usado indebidamente.
about this problem for a long time,
de preocupación hace tiempo,
al mercado por primera vez.
first hit the market.
on countermeasure technology,
en tecnología de prevención,
effort at AI Foundation,
of machine learning and human moderators
automático con el control humano
combatiendo mi propio trabajo.
lanzar se llama "Reality Defender",
is called Reality Defender,
that can flag potentially fake content
que detecta contenido potencialmente falso
causar un gran daño,
that we make everyone aware
esté al tanto de este tipo de cosas
and be critical about what we see.
correctas y juzgar lo que ve.
we can fully model individual people
completamente una persona en particular
the safety of this technology.
de esta tecnología.
y lleno de esperanza,
se usa con buen juicio
positive impact on the world
persona en el mundo
el futuro que queremos.
the way we want it to be.
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