question about<Semantic Brightness Consistency Constraint loss>code
Opened this issue · 2 comments
JetsonKarl commented
Myloss.py line143:d2 = torch.mean(a2)
JetsonKarl commented
it can not represent equation 6. There are some zeros in the tensor but you have not deleted them. So it can not represent this area's average brightness level, the truly B is bigger than your result.
LingLIx commented
它不能表示方程 6。张量中有一些零,但您没有删除它们。所以它不能代表这个区域的平均亮度水平,真正的B大于你的结果。
Yes. You are correct. We will change this part of the code later. Thank you very much for asking this question.