h a l f b a k e r y"Bun is such a sad word, is it not?" -- Watt, "Waiting for Godot"
add, search, annotate, link, view, overview, recent, by name, random
news, help, about, links, report a problem
browse anonymously,
or get an account
and write.
register,
|
|
|
Please log in.
Before you can vote, you need to register.
Please log in or create an account.
|
I google image search people a lot and work in the blind
community where few people have their picture online, and
I have noticed that the average of an image search for
their name usually gives you an approximation of what
they look like anyway, whether this is because of the
ethnic history
of names or association with friends who
look like you I am not sure.
So this is a virtual entity created at first by averaging all
of the images in a Google image search for instance of a
persons name, with an algorithm that focuses on faces and
decides whether something is an anomaly or not etc. This
could eventually include videos and other stuff, writing etc
to create an avatar/virtual self.
Friends with Benefits
http://phys.org/new...c-similarities.html
[JesusHChrist, Jul 14 2014]
[link]
|
| |
Is this intended to use last names, first names, or full
names? |
|
| |
Because I'm pretty sure the first wouldn't produce a
close match in most cases, the last might, for some
names, and full simply won't find enough matches in
many cases to average. |
|
| |
Maybe you could put into the algorithm to use
whatever worked the most predictably, so if
someone had a full name that was very exclusive to
them and they were not online you could reduce it
to their last name, and if someone had a really
common name you could add in key words from any
other information you had about them. |
|
| |
If
you happen to be named John Smith, unless you
look
like Captain John Smith, or the animated Disney
version of the same, you're not going to have
much
luck with a match. |
|
| |
I think this is going to have severe problems with
most common or very unique names. It might
work for a certain subset of the middle ground,
where most individuals who have it are closely
related. |
|
| |
For instance, people with my last name tend to be
relatively closely related to me, and my first name
is not unusual in the family, but the image search
turns up (besides lots of gravestones) people of
differing genders and races about as frequently as
it turns up actual relations. |
|
| |
It also turns up a lot marionettes
(I actually know why, distant historical relation,
with a similar not identical name was known for
them), that would probably get
averaged in, producing really odd results. |
|
| |