search

swz30 / Restormer

[CVPR 2022--Oral] Restormer: Efficient Transformer for High-Resolution Image Restoration. SOTA for motion deblurring, image deraining, denoising (Gaussian/real data), and defocus deblurring.
Other
1.65k stars 220 forks source link

train patch #53

Open 123456789-qwer opened 2 years ago

123456789-qwer commented 2 years ago

When performing defocus blur training, do you pre-crop the image to 10 patches？

adityac8 commented 2 years ago

Hi For the Defocus Deblurring task, we extract patches from the DPDD dataset. The details can be found here Thanks

RuiWang-coder commented 2 years ago

I have some question about the DPD Dataset used in Restormer, did you guys change the Dataset? I thought the dataset archs was different from the original DPDD, can you tell me what changes did you do compare to original DPDD? Looking forward to your reply!

c-yn commented 2 years ago

@RuiWang-coder It's the same for the motion deblurring task. They pre-crop the image into several parts.

adityac8 commented 1 year ago

We do not make any changes to the original dataset for images. We only change the directory structure to suit our preprocessing.