Genetic Algorithms: Evolving a human face

[JDay373]

what is with all these fake replies?

[thehaircrow]

cool idea

[Mjhond]

I am very happy to see the vidoe after you give this o one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[Ondelendo]

I Love The Video It Can Increase My Knowledge so one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[anakmudajaman]

Steady I Really Like This Video so one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[bebeheuy]

Good, I like that you share this video, I wish success always so one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[willamricard]

I Really Like The Video From Your so one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[bundawartini]

Your Video Is Very Useful Sharing so one friday afternoon I got the crazy idea of creating a genetic algorithm to evolve human faces using our detector as the fitness function.

[TiagoTiagoT]

And about the local maximum/minimum; perhaps you could make it so the frequency and/or intensity of mutations and crossovers is inversely proportional to the difference between the best scores of this generation and the previous' ?

Perhaps with a delay or moving/weighted average instead of just using the current difference from the last, to reduce feedback loops leading into extreme oscillations, and smooth out the changes in intensity

[TiagoTiagoT]

Btw, for doing the crossover; wouldn't it be just a matter of picking two of the best individuals of each generation, and then doing it like you do with plain mutation but instead of being random it's the average of the color for the pixel in the two parents (probably you wanna have each mating produce at least two new individuals, one is parent A with some pixels mated with parent B's and the other is parent B with the same pixels mated with parent A's) ?

[TiagoTiagoT]

An interesting approach to see what the detector thinks a face should look like.

[justinlloyd3]

Why not crowd source it and use a human choice as the fitness function?

[PhantomAct]

Isn't it weird that hot chicks in dreams don't have faces or barely recognisable ones?

[taostoner1]

haha, thats freakin awesome! So the face recognizer will return a value that represents a percentage match for the face? and they evolve to get the highest match? really neat idea.

[ImaginaryMdA]

Wow that is really impressive, given it doesn't have anything to copy or ressemble!

Could you perhaps use this to improve your face recognition software?

In sort of a predator-prey simultaneous evolution,

the fake faces are the prey and the software is the predator,

There also have to be some real faces with the fake ones, these can be 'poisonous food' and to complete the analogy 'eating' is recognising as a fake.

This seems to work in nature as Mimicry.

[DavidMTRutten]

What did you use as the 'genes' for these images?

[cirusMEDIA]

I think i saw a DOG in there for a split second !!!

[SalsaTiger83]

one lesson from this if you don't want your face recognized by this particular face recognizer: stop smiling!

[MacHans78]

A very creative idea! Nice!

[redtihor1]

Can we do something like this in color;You're probably using fitness for two parameters;Won't fitness for rgb and brightness increase the time required by nearly 32 times?

[joecougar67]

looks like zach galifianakis

[ernesto50]

congratulations, it seems to get stuck in a local minimum

[lnpkural]

Same pictures used in cycles..

But its good if there is some Genetic Algorithm behind it. AI .

[Y0ts0]

so beautiful. not.

[quietthomas]

So this is what the machine knows of us - and hence, this is the machine's chosen face. A happy, chubby, internationalist. I like it.

[Ulterior1980]

Hmm, a random data gets randomized until found fit - not impressive

[Pixelgel]

@Ulterior1980 Try learning what the program is actually doing before making a fool out of yourself.

[Ulterior1980]

@Pixelgel You dont understand how the fitting algorithm works, and then you made the fool of yourself

[Ulterior1980]

@Pixelgel Ever heard of a term confussion matrix, mor.on?

[michielvda]

Its freaky when we put computers in this sort of dream-state. Remind me like the bloke who does the 'deep-learning' thing for feature extraction in character sets.

Sometimes this face here is so damn haunting... no wonder us humans have nightmares when our brain gets to run free at sleep-time.

[icarus119]

seeing this will give idiots the wrong idea. how good it would have been if all the people in the world were smart enough.

[jamesarongray]

IT'S GOD!!!!

[moseskiiza]

looks like that face on mars

[lardhat]

One of the coolest ideas i've ever seen.

[b4ra2]

should have added music

[Houshalter]

Whats the fitness function, it seems its to forgiving to failures that doen't even look close to a face. But you did succeed in making one of the creepiest videos on youtube. Im going to have nightmares about that face.

[stanbwonderful]

Amazing!

[kjlg74]

That's pretty cool!

[pahilario]

it shows that you an effective face detector.

[djJoz]

How can I get a license of the FTR SDK?? How much does it cost?

[ludamad]

It's evolving against their face recognizer, I guess it has a measure of likeness to a face used as the fitness function.

[ravivyatom]

Very Nice !

[bja6a]

Looks like the face on Mars. Whose face is that?

[sleepingeye]

John Doe! ;)

[Dirtfire]

Since that's a totally generic face, I think it's neither male nor female --or it's both.

[sleepingeye]

right