search

ewrfcas / MAE-FAR

Codes of Learning Prior Feature and Attention Enhanced Image Inpainting (ECCV2022)
Other
78 stars 4 forks source link

About resolution #15

Closed UdonDa closed 1 year ago

UdonDa commented 1 year ago

Hi, @ewrfcas

Thank you for providing your nice implementation! I have a question.

You evaluated your model using a 512x512 resolution. However, you provide only 256x256 weight using a Places dataset. Why don't you provide 512x512 weights?

ewrfcas commented 1 year ago

The released model has been finetuned on 512x512 images.

UdonDa commented 1 year ago

The input for MAE is 256x256, but the rest network is 512x512, right? Please excuse my lack of understanding.

ewrfcas commented 1 year ago

Yes, the feature will be resized as mentioned in the paper.

UdonDa commented 1 year ago

I understand your method perfectly. Thank you for explaining to me!! I can run your code.

Thank you.