CRFLoss for regression - Githubissues

meng-tang / rloss

Regularized Losses (rloss) for Weakly-supervised CNN Segmentation

MIT License

207 stars 47 forks source link

CRFLoss for regression #18

Open Tord-Zhang opened 3 years ago

Tord-Zhang commented 3 years ago

Hi, since the log probability is involved in the crf loss, I suppose that it can only be used for classification task? What about regression task? is there any good solution?