Skip to content
New issue

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

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

Already on GitHub? # to your account

predicted mask is whole black #38

Open
mxy519 opened this issue Nov 5, 2020 · 5 comments
Open

predicted mask is whole black #38

mxy519 opened this issue Nov 5, 2020 · 5 comments

Comments

@mxy519
Copy link

mxy519 commented Nov 5, 2020

There is no problem in training and testing process ,but for each image, the ac is 100, miou is 50, and the predicted mask is whole black, what shoud i do?

@Tramac
Copy link
Owner

Tramac commented Nov 13, 2020

Hello, you can check if the label is right.

@mxy519
Copy link
Author

mxy519 commented Nov 13, 2020

The label is right. There are two classes(background and iris) in my dataset, and the input images are one channel ,also,the gt label is one channel. I do not know where is wrong

@Tramac
Copy link
Owner

Tramac commented Nov 13, 2020

Maybe you need to check the pixel value of the predicted mask.

@mxy519
Copy link
Author

mxy519 commented Nov 13, 2020

Thanks, I will do it. There is another problem,for each image, the ac is 100, miou is 50, is it normal?

@Tramac
Copy link
Owner

Tramac commented Nov 13, 2020

Probably not.

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

No branches or pull requests

2 participants