8000 The reproduction for semantic segmentation · Issue #19 · yzd-v/MGD · GitHub
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

The reproduction for semantic segmentation #19

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
Open
smalllwater opened this issue Apr 2, 2025 · 0 comments
Open

The reproduction for semantic segmentation #19

smalllwater opened this issue Apr 2, 2025 · 0 comments

Comments

@smalllwater
Copy link

I have been replicating your work MGD recently. First of all, thank you for your contribution to mask knowledge distillation, your work has been a great inspiration to me.
During the reproduction process, I tried to keep the environment configuration consistent with the original configuration, applying the same torch, torchvision, mmcv, mmseg versions. Also tried to change various parameters, including learning rate, BN ,init_student , use_logit. I also tried adjusting the pretrained parameter to init in backbone, or downloading pre-trained weights to a local device. However, no matter how I adjust, my replication results are still well below expectations, with a 10% mIoU gap.
Would you please give me some guidance or advice?

20250401_163951.log
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
1 participant
@smalllwater
0