Question about the setting in your experiment about ccvc HOT 4 CLOSED

Hi,
1 & 3. The experimental settings of our backbone are consistent with UniMatch, maybe you can change the dilation settings and carry on some experiments. Unfortunately, we did not perform such experiments before.
2. Yes, there are 2 labeled and 2 unlabeled data on one GPU and there are 4 GPUs in total.

from ccvc.

Thanks for your answer! In question 1 I mean the dilation in the code of Cityscapes is set as [12,24,36] and the replace_stride_with_dilation is [False, True, True], which are not consistent with UniMatch. So I don't know whether it is the final setting of your experiment.

from ccvc.

Ah, I see, UniMatch modifies the settings, I adopt the dilation settings from the last version of UniMatch. Yes, This is the final setting of my experiment ([False, True, True]).

from ccvc.

Get it. Thanks!

from ccvc.