Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

损失函数的参数好像有点问题 #37

Open
zust-pss opened this issue Oct 26, 2021 · 4 comments
Open

损失函数的参数好像有点问题 #37

zust-pss opened this issue Oct 26, 2021 · 4 comments

Comments

@zust-pss
Copy link

(self.l_coord * loc_loss + 2 * contain_loss + not_contain_loss + self.l_noobj * nooobj_loss + class_loss) / N

contain_loss前系数为何是2?
not_contain_loss前应该有self.l_noobj作为系数

@punk1
Copy link

punk1 commented Mar 18, 2022

请问明白了么?我刚好也在纠结 为何前面系数 2

@myshrry
Copy link

myshrry commented Jun 23, 2022

按照论文应该没有2的,还有就是为啥有not_contain_loss这个东西,不就是nooobj_loss吗?

@Vandaci
Copy link

Vandaci commented Jul 20, 2022

作者的loss跟论文不一样,我建议如果要复现的话,可以自己重新写一下

@xiazhi1
Copy link

xiazhi1 commented Mar 5, 2024

我认为应该是作者在实际调参的时候通过测试和训练发现模型预测出来的置信度过低或contain loss过高 所以调整了权重参数以获得更好的检测效果

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

5 participants