1
$\begingroup$

I have some dataset ${(x1,y1), (x2,y2)...(xn,yn)}$, where, $x$ is the picture of a facial expression，while $y$ is the fraction corresponding to their degree of the happiness (happy laugh: $y$ close to $100$; sad cry : $y$ close to $0$). The range of the $y$ is $0$~$100$.

In my dataset, I didn’t have the sample with $y$ value greater $80$，but I want to get a very happy facial expression. In other words，I want to get a facial expression with very high y value $(y>95)$.

Can GAN (Generative Adversarial Networks) implement this? If not，is there any model that can implement this goal?

Improve this question
$\endgroup$

1 Answer 1

Reset to default
1
$\begingroup$

Here's a paper by Tom White and its repository studying the specifics of sampling from generative models. Section 3 in the paper, Attribute Vectors, might be of interest for your problem. I think you'll likely have to rely on another dataset to learn the latent code for a smile to use on your own images, but given the results of the author's own work it should be fairly good.

Hope this helps!

Improve this answer
$\endgroup$

Start asking to get answers

Find the answer to your question by asking.

Ask question

Explore related questions

See similar questions with these tags.