Hello, may I ask if vig can be used in semantic segmentation? If so, where should it be modified?
I tried to modify only the prediction layer, but found it difficult to restore the dimensions [1,n_filters,1,1] output before the prediction layer to the original resolution [1,classes,H,W]
Hello, may I ask if vig can be used in semantic segmentation? If so, where should it be modified?
I tried to modify only the prediction layer, but found it difficult to restore the dimensions [1,n_filters,1,1] output before the prediction layer to the original resolution [1,classes,H,W]