xinntao / SFTGAN

CVPR18 - Recovering Realistic Texture in Image Super-resolution by Deep Spatial Feature Transform

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

"manually change the probability map to each category"

jiequanz opened this issue · comments

What does "manually change the probability map to each category for a certain input" mean in the supplementary materials?

Here is my understanding, but I want to make sure.
Before changes, for one picture, there are 8 category maps (sky, water, building, etc). If we want a building prior only, we just use the "building" channel to replace the others, so now we have 8 same category maps(all buildings). Is that right?

Each channel represents the probability of its category. If we want to change to building prior, we set the corresponding (like building) change with prob =1, and the left channels = 0.

Thanks for your reply!
I want to follow up a little to make sure I understand. Appreciate your help!
image

One channel is a map that has many colors. Each color represents different probability. Let's say that yellow represents high probability and blue represents low probability. What you said is to make the "building" map all yellow, and the rest all blue. Is that right?

Yes, right.