1
$\begingroup$

I want to help people with cancer who are under chemotherapy, and generally people who have lost their hair to Virtually Try-On Toupees/Wigs on their head.

VTO must support both the frontal and side positions of the head.

At first, I thought I could use traditional deep-learning to find landmarks, then place the hairstyle on the head.

But the results were unrealistic and inaccurate.

2D Placing of Hair:

Some Hairstyles and Portraits fit well together:

enter image description here

But some don't:

enter image description here

They need manual modification:

enter image description here

Sometimes the original hair needs to be erased:

enter image description here

GANs promise more natural results with less manual engineering

changing hairstyle using GAN: enter image description here

changing hairstyle using GAN: enter image description here

So, I decided to use GANs for this Task.

Is it a good choice or is there an alternate solution?

$\endgroup$
5
  • $\begingroup$ You really need to provide more details. What are you using the GAN for? Moreover, try to ask a question that will not lead to opinions. Edit your post to clarify that. $\endgroup$ – nbro Dec 20 '20 at 20:04
  • $\begingroup$ @nbro I edited mu question, I don't know if its enough or not? $\endgroup$ – aliep Dec 20 '20 at 20:30
  • $\begingroup$ Where does this conclusion "GANs promise more natural results with less manual engineering" come from? $\endgroup$ – nbro Dec 20 '20 at 20:38
  • $\begingroup$ @nbro because a well architected and well trained GAN tends to generate realistic fakes, but the traditional approach of placing a transparent hairstyle on the head, needs a lot of refining to the both hairstyle and the portrait itself $\endgroup$ – aliep Dec 20 '20 at 20:53
  • $\begingroup$ Have you tried to search for research work that attempts to do similar things? Maybe someone has already tried to do that. Feel free to edit your post to include more info. $\endgroup$ – nbro Dec 20 '20 at 20:54

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.