chainer-fast-neuralstyle icon indicating copy to clipboard operation
chainer-fast-neuralstyle copied to clipboard

Published 20 hours ago • verified publisher icon yusuketomoto

Some problem about your re-impletation

Open DavieHR opened this issue 8 years ago • 5 comments

GOOD WORK. But I have some Questions to ask you. First, I used your what you have said in the comment to train a new model, but I still found that the results are noisy. Second I found that you just subtracted 120 per image, why did not subtract the mean value that VGG provided?

DavieHR avatar May 19 '16 09:05 DavieHR

Yeah. I have the same question to ask.

eric-xw avatar Jun 30 '16 21:06 eric-xw

@DavieHR @littlekobe I think that doesn't make a lot of diference. This VGG mean is just around 120.

austingg avatar Jul 15 '16 07:07 austingg

Excuse me, I found that both y and xc are subtracted by 120, but the style imge is not. Is there anything wrong when calculate L_style with feature_hat(subtracted by 120) and feature_s(not subtracted)? @austingg @DavieHR

yanhn avatar Aug 17 '16 10:08 yanhn

@yanhn you subtract 120 just before you feed it to VGG network. For the style it's subtracted by vgg.preprocess method :)

cryptexis avatar Aug 17 '16 13:08 cryptexis

@cryptexis Thank you. Yes，you're right. My bad, I didn't read code carefully.

yanhn avatar Aug 18 '16 01:08 yanhn