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

Will the proposed datasets be released ? #6

Open
Zeqiang-Lai opened this issue Sep 12, 2023 · 3 comments
Open

Will the proposed datasets be released ? #6

Zeqiang-Lai opened this issue Sep 12, 2023 · 3 comments

Comments

@Zeqiang-Lai
Copy link

  1. Image Editing in the Wild (IEIW),
  2. Object replacement based on SA-1B, Open Images
  3. PhraseCut with LAMA
  4. Web crawl.
@Gengzigang
Copy link
Collaborator

Thank you for your interest in the proposed datasets. Due to some copyright concerns, we currently have no plans to release these specific datasets. However, we believe that the publicly available datasets should be sufficient for training a model with similar performance to ours. We appreciate your understanding.

@Zeqiang-Lai
Copy link
Author

Thanks for the rely, could you share the gpu budgets for constructing these datasets ?

@Gengzigang
Copy link
Collaborator

If you would like to train on all tasks (955k images per epoch) simultaneously, it took us 3.5 days to train for 200 epochs using 48 V100 GPUs. However, if removing IEIW or focusing on training specific tasks or datasets, our method is fully adaptable, and the training time will be significantly reduced.

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

2 participants