wickwick
wickwick
I think CBAM is a cascade of channel attention and spatial attention , and BAM is a parallel connection of this two modules
> We directly use the pre-trained models to run the results for fair comparisons. For fair comparisons, all the 13 methods should be trained using the same dataset. Isn't it?
> I mean we directly use the models released by the corresponding authors. We did not re-train these models as we are not sure whether or not the retrained model...