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

rpn foreground accuracy is always 0 #81

Open
chengweige opened this issue Mar 6, 2019 · 3 comments
Open

rpn foreground accuracy is always 0 #81

chengweige opened this issue Mar 6, 2019 · 3 comments

Comments

@chengweige
Copy link

i trained the alex-9s-600-rpn-base without finetune,here is the snapshort result
default
but the rpn foreground accuracy is always 0, is it alright?

@zhaoweicai
Copy link
Owner

It seems not right. Try finetuning first, and compare the difference. I don't think training-from-scratch works if you don't make any change, e.g. hyper-parameters.

@chengweige
Copy link
Author

Thank you for your reply. I have solved the problem.

@xiaosibai
Copy link

Hi!The IOU always equals -1 when i use this "res101-9s-600-rfcn-cascade" to train my data ,can you tell me the reason!? And how to get the value of anchors? The loss ( floating up and down at 0.5)won‘t drop no mater how i try? Those problems are really killing me now, I cann't figure it out, can you help me?1

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

3 participants