The result of cityscapes is strange

[canglangzhige]

Thanks for your work.
When I train the model with Cityscapes follow https://github.com/dvlab-research/PanopticFCN/issues/21. I get the result like:

But the ground truth is:

You can see that the segmentation of the front end at the bottom of the image(inside the red rectangular box) is chaotic.
Why is that?

[yanwei-li]

Hi, this is the background class that is not included in the evaluation. It is chaotic because we do not take this class in training, which causes the random prediction. You can take this ignore class into training GT if needed.