Closed yz-ju closed 5 months ago
Hi, sorry that I can't find the code for that. Anyway, it's easy to code:
outs_fused
where outs_fused.append(outs_rgb/x[i])
each time. This means we discard the feature fusion operations and send the feature extracted from self.vssm directly to the decoder.
There are unimodal segmentation results in the paper. How to set up and implement unimodality?