What's the meaning of sample_reg_loss?
exiawsh opened this issue · comments
Hello! Thanks for the great work!
I'm wondering the meaning of sample_reg_loss. Was it supervised during training?
I am not certain since the sample_reg_loss
is not described in the paper, but I think it's for training scoring net directly and makes the score sparse. But I still don't know why this loss can help scorenet to score foreground-related feature vectors with high scores.
Any advice? @twangnh