Can the model realize monocular depth estimation?

[ZhouMengjie]

Hi, thanks very for your brilliant work. I would like to output the dense feature map and point map with your work only for one image. But I found when I input two same images to each branch, I got very strange feature map (dense local features) but correct point map. Could I know whether your model can be used to process same inputs? And whether you have checked the 24-D feature map for the visualization?