About the feature-level visualization

[manmanjun]

Hi,
Thanks for your good work. I have a question about Figure3 in the paper: what is the way you do the feature level visualization.
Thank you.

[Dootmaan]

Hi there, sry for the long time waiting. I have been working on another project so i didn't pay much attention on this repo recent days.

I have to clearify that im not one of the authors of the paper and this is not the official implementation of method proposed in Dual Super-Resolution Learning for Semantic Segmentation. I was also not sure how they generate Fig.3. Personally i think they just project the features onto a 3-channel space and visualize it as an image.