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

Loss is nan #35

Closed
994374821 opened this issue Aug 6, 2021 · 4 comments
Closed

Loss is nan #35

994374821 opened this issue Aug 6, 2021 · 4 comments
Labels
individual Individual problem and need

Comments

@994374821
Copy link

994374821 commented Aug 6, 2021

作者,您好!

当我使用自己数据集测试您代码时,l_det_loc、L_det损失是 nan,l_imgcls一直是0

自己数据集只有行人类,已经修改过datasets/voc.py中的CLASSES=('person',)

训练时使用的1个显卡

@yuantn
Copy link
Owner

yuantn commented Aug 6, 2021

你好,请参见问题 #24,看看能否解决你的问题。


Hello, please refer to Issue #24, and see if it can solve your problem.

@994374821 994374821 reopened this Aug 6, 2021
@994374821
Copy link
Author

将模型类别改为2(实际GT只有一类),l_imgcls正常了,但是l_det_loc、L_det损失依旧是 nan

@yuantn
Copy link
Owner

yuantn commented Aug 6, 2021

你可以看看你的数据集本身的边界框标注信息是否有问题,或者尝试调整学习率,并重新做一下实验,看看实验结果。


You can see if there is any problem with the bounding box annotation of your dataset, or try to adjust the learning rate, and re-run the experiment to check the experiment results.

@yuantn yuantn added the individual Individual problem and need label Aug 6, 2021
@994374821
Copy link
Author

已解决,排查是因为标签问题

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

No branches or pull requests

2 participants