search

leehomyc / Faster-High-Res-Neural-Inpainting

High-Resolution Image Inpainting using Multi-Scale Neural Patch Synthesis
http://www.harryyang.org/inpainting
MIT License
1.3k stars 213 forks source link

Unknown image x in the paper #33

Open JieLiu95 opened 6 years ago

JieLiu95 commented 6 years ago

Hello, thanks for your work. In Content Network, we will use holistic content loss Ec(h(x, R), f(x0)), but how we get the unknown image x? And what the relationship between x and f(x0), does it have a function about x and f(x0)? What kind of difference between f(x0) and h(x, R)? Thanks!

cmyyy commented 5 years ago

@JieLiu95 ，hello！I have the same question. Have you figured it out？

JieLiu95 commented 5 years ago

@cmyyy , In holistic content loss E_c(h(x, R), f(x_0)), x is the ground truth.

cmyyy commented 5 years ago

@JieLiu95 ，then how about the x in local textual loss？