what colormap is used for depth completion and its range value ?
Opened this issue · 3 comments
Hi, recently I went through your great work and I would like to know what colormap is used for depth completion and its range value means every color in the colormap is referred to some error value, can you tell me the error range value of the color map?
And I appreciate that in your depth completion, you rendered with different colors where there is no valid ground truth point or valid depth point, most of the methods won't do that, I am wondering how to do that! Thanks in advance!
Hi, The visualization in the README file are actually the depth values, not the error values. I think I used the default color map of matplotlib.
I don’t understand what do you mean by “rendered with different colors where there is no valid ground truth point or valid depth point”, could you detail a bit more?
Hi, jiangwei221 sorry for the undetailed question, My question is, we usually compute errors for predicted depth maps with help of ground truth. But in-ground truth in some pixel, we don't have depth value as you see in the picture (ground truth) for far away distance then how in prediction(see in the picture) we are getting depth value which is in yellow color where there is no ground truth.
For every color in the colormap, it must have a range value like for example 0-10 m it represents in purple color and 10m -20m it represents in blue color. So in that case, what error range value does colormap represents?
Hi Laihu08, I think matplotlib automatically normalize the image with max/min values. I didn’t do extra clamping or normalization(if I remember correctly), so I’m not sure about the depth range of each pixel color represents.