Inconsistent results in Table 3 (SYNTHIA->Cityscapes)

[BinhuiXie]

Hello, thank for sharing the great work.
I find some inconsistent results in Table 3. Why the reported results of the CAG-UDA model in two mIoU metrics (13- and 16-class subsets) are different? Previous SOTA methods, like AdvEnt, report the results on the 13-class subset based on that of 16-class subset. Could you please explain it?

from [arxiv](https://arxiv.org/abs/1910.13049)

[RogerZhangzz]

Hi,
Thanks for your interest. 'IoU' means intersection over the union. When the class number decreases from 16 to 13, the redundant 3 classes should be ignored, leading to the denominator decrease during computing IoU. Thus the reported results on each class are inconsistent.

[TianX1nYu]

Hello, thank for sharing the great work.
I have a puzzled question for you. Is syhthia's warmup stage training conducted in 19 categories? Are 19 categories still used in the next three stages?
PS: can you share the model weight of Synthia dataset that you work in CAG? Thank you very much.