Here's something a lot of people still don't know: The latest DeOldify doesn't use GANs anymore. And I'm not being cute with terminology- NoGAN isn't used either. We needed something more production worthy and controllable and it just wasn't cutting it. 1/
-
Show this thread
-
It's been a year since we last used GANs. You may have been lead to believe that GANs are -the- way to get realistic results but believe me, there's actually better ways IMHO. I can't tell you exactly what we're doing now but I can tell you this: 2/
4 replies 2 retweets 72 likesShow this thread -
It was a good thing I got past my ego (after some time) and listened to
@jeremyphoward and@fastdotai when they said they that they are getting "better than GAN" results using perceptual loss in super resolution :) 3/9 replies 11 retweets 174 likesShow this thread -
GAN generates details in its way. I feel like the result with GAN is better but quantitatively worse. I usually consider perceptual loss firstly rather than GAN. Since plausible colors can be different from the ground-truth, how did you evaluate your results?
2 replies 0 retweets 0 likes -
So we have two metrics: FID (super slow) and another one that's very fast that isn't quite as good but gives us fast feedback across resolutions to let us know how good the generalization on that is. Unfortunately neither of these are 100%, so visual inspection is a must. Lots.
2 replies 0 retweets 5 likes -
What's FID?
2 replies 0 retweets 1 like -
Oh just Frechet Inception Distance
1 reply 0 retweets 3 likes -
Replying to @citnaj @jeremyphoward and
Honestly I’ve found that when evaluating a GAN, nothing beats generating and manually reviewing images at every n steps.
1 reply 0 retweets 4 likes
That's what we had to do for NoGAN- find a stopping point based purely on visual inspection. And it was very fine grained- within 100-200 iterations- where you had to look! That was the attempt to "control" it. Makes it extremely hard to make repeatable experiments.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.