Can the rmi_loss part be directly applied to two natural images, instead of segmentation labels, to calculate the mutual information loss? - Githubissues

RElbers / region-mutual-information-pytorch

PyTorch implementation of the Region Mutual Information Loss for Semantic Segmentation.

MIT License

25 stars 1 forks source link

Can the rmi_loss part be directly applied to two natural images, instead of segmentation labels, to calculate the mutual information loss? #6

Open zhongxinyun opened 2 months ago

zhongxinyun commented 2 months ago

Can the rmi_loss part be directly applied to two natural images, instead of segmentation labels, to calculate the mutual information loss?