liheyoung / freemask Goto Github PK

When I ran filter.py, I encountered this error. RuntimeError: CUDA error: an illegal memory access was encountered

Hi @LiheYoung,

I received an error when trying to download the ADE20K_Synthetic.zip file from OneDrive SharePoint.
The error message stated:

The file size exceeds the allowed limit.
CorrelationId: b3dc19da-f990-432e-95f5-5e3e26e50e35,
UTC DateTime: 11/08/2023 13:35:48

Would you be able to split the dataset into smaller collections and re-upload them?

Many thanks

Hello, would you have plan to release the code of filtering criterion and sampling principle, Thanks!

Thanks the authors for the inspiring work! It has contributed to my understanding and research in this field.
I am also curious if the authors have attempted on datasets like PASCAL VOC2012?
Look forward to any further information you provide.

Hi. Since the FreestyleNet takes text and layout as input, what are the prompts that you used for generating data?
How to automatically set different prompts for multiple generation?

Hi, author! Thanks for your great work. I have a question about how to achieve domain-specific semantic image synthesis. For example, I want to use a semantic mask from Cityscapes and an image from GTA5 as style references to generate a GTA5-styled image in an unsupervised domain adaptation (UDA) task. Compared with FreeMask, I add an image for style references. Do you have any ideas about how to deal with this additional image to obtain style information for FreeStyleNet? Looking forward to your reply.