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

有人成功复现嘛?效果是多少? #29

Open
zhixpc opened this issue Sep 8, 2023 · 4 comments
Open

有人成功复现嘛?效果是多少? #29

zhixpc opened this issue Sep 8, 2023 · 4 comments

Comments

@zhixpc
Copy link

zhixpc commented Sep 8, 2023

我复现了之后,改了代码,因为只有一个gpu,不能用分布式,打算跑800epoch,跑了200epoch,miou效果还是只有0.多,所以想问有没有人能跑出原文的效果的?

@zhixpc
Copy link
Author

zhixpc commented Sep 8, 2023

对了,我的batchsize是2

@T-SHAN
Copy link

T-SHAN commented Sep 8, 2023

我有个问题,原论文中使用resnet50时miou为50.4%,后边又说用resnet50跑出结果为51.3%。这个51.3%时通过多尺度验证得出的吗?同理使用resnet101跑出结果52.4%,这个也是用多尺度验证得出的吗?

@zhixpc
Copy link
Author

zhixpc commented Sep 9, 2023

我有个问题,原论文中使用resnet50时miou为50.4%,后边又说用resnet50跑出结果为51.3%。这个51.3%时通过多尺度验证得出的吗?同理使用resnet101跑出结果52.4%,这个也是用多尺度验证得出的吗?
有可能是因为输入尺寸,50.4的时候他写的输入3480480,51.3的时候没有说,可能是原尺寸

@ChrisEvans2
Copy link

你的batch-size太小了

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