uncertain regions Loss parameter sigma
jason-xiao-yi opened this issue · 6 comments
Hello, author. May I ask how the sigma parameter in the loss of uncertain region in the paper was obtained? Is it to be learned as a learnable parameter, or is it to be learned as a distribution factor of a deterministic region? In the final loss function, adding logsigma to each pixel would cause the second term to be too large
Hello, author. May I ask how the sigma parameter in the loss of uncertain region in the paper was obtained? Is it to be learned as a learnable parameter, or is it to be learned as a distribution factor of a deterministic region? In the final loss function, adding logsigma to each pixel would cause the second term to be too large
give me your e-mail, I send the code to you.
Dear author. Compared with other SOD tasks, the method of uncertainty loss made a great improvement on the blur edge process(especially on Bμ and BDE) may I see some details about the uncertainty loss? My e-mail is: 1140395492@qq.com, thank you so much.
Dear author. I have read the article and found the uncertainty loss really made a great improvement on the boundary of some objects. But I didn't understand how can I get the sigma in that loss. Could you please send me the code? E-mail: 1047348834@qq.com . Thank you.
Could you please send me the code of uncertainty loss? E-mail: shkarupa.alex@gmail.com . Thanks.
Could you please send me the code of uncertainty loss? E-mail: 1072593143@qq.com . Thanks.
Could you please send me the code of uncertainty loss? E-mail: eric3772443@gmail.com . Thanks.