roimehrez
commented
6 years ago What you suggested is exactly what the authors of SRGAN did. The perceptual quality is a bit less good.
בתאריך יום ו׳, 15 ביוני 2018, 17:09, מאת zhangmozhe < notifications@github.com>:
Hi I really like the paper. One question is that in your super-resolution paper you optimize the contextual loss with conv3_4 features. How about directly minimizing the perceptual loss for the conv3_4 features? What's the visual difference. Thanks!
— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub https://github.com/roimehrez/contextualLoss/issues/8, or mute the thread https://github.com/notifications/unsubscribe-auth/AOoKJtOArzL1ivWJHqVmHPNJMaYlMWkdks5t88AKgaJpZM4Upn6q .
Hi I really like the paper. One question is that in your super-resolution paper you optimize the contextual loss with conv3_4 features. How about directly minimizing the perceptual loss for the conv3_4 features? What's the visual difference. Thanks!