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.

Sign up for GitHub

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

Jump to bottom

请问pascal voc是加载官方预训练跑出来的？ #10

Open
joe3154 opened this issue Mar 12, 2023 · 13 comments
Open

请问pascal voc是加载官方预训练跑出来的？ #10

joe3154 opened this issue Mar 12, 2023 · 13 comments

Comments

@joe3154
Copy link

joe3154 commented Mar 12, 2023

我自己跑官方的预训练backbone效果很差。
@bubbliiiing
Copy link
Owner

是的，pretrained=True吗

@worldlyesterday
Copy link

我也难以复现原作中ADE20k数据集的mIoU，可能需要调整超参数？

@joe3154
Copy link
Author

joe3154 commented Mar 13, 2023

我参数配置调的都是一样的，用的ADE预训练权重，但是跑voc很差

@worldlyesterday
Copy link

请问一下你是怎么载入官方的预训练权重呀，我之前试了一下匹配不上

@joe3154
Copy link
Author

joe3154 commented Mar 13, 2023

dict按key加载

@joe3154
Copy link
Author

joe3154 commented Mar 13, 2023

不冻住backbone效果好差

@bubbliiiing
Copy link
Owner

0 0不应该啊，你们是怎么操作的，我就是正常的pretrained=True开始的
不是官方那个开始的

@bubbliiiing
Copy link
Owner

你们的训练参数啥是什么

@joe3154
Copy link
Author

joe3154 commented Mar 14, 2023

除了LR_Scheduler不是poly，都和官方一样

@joe3154
Copy link
Author

joe3154 commented Mar 14, 2023

我加载你voc的b2可以跑到78.38%，只加载imagenet跑50epoch一半都不到。

@worldlyesterday
Copy link

我加载voc的b2预训练权重在ADE20k上训练，和原论文的实验结果也有挺大差距的

@joe3154
Copy link
Author

joe3154 commented Mar 14, 2023

非常感谢！ @bubbliiiing ，没有问题了。

@kimer1
Copy link

kimer1 commented Sep 9, 2023

请问怎么解决跑ADE20k数据集效果很差的问题

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
4 participants
@joe3154 @bubbliiiing @worldlyesterday @kimer1