search

lixinustc / Causal-IR-DIL

Distortion invariant feature learning for image restoration from a causality perspective
Apache License 2.0
43 stars 2 forks source link

Confusion about cross-distortion types #3

Closed chensming closed 1 year ago

chensming commented 1 year ago

Hi, you have done a nice job and I am interested in it. When refer to "for cross-distortion types, DILps achieves better performance for RealSR and Image Deraining", I'm confused as to why a different dataset is a different degenerate type (eg. raining datasets).

lixinustc commented 1 year ago

Thanks for your questions. For deraining, the ways of producing the rain in different datasets are different, such as rain stick or raindrops, dense or sparse rain, etc.

chensming commented 1 year ago

Thanks for your questions. For deraining, the ways of producing the rain in different datasets are different, such as rain stick or raindrops, dense or sparse rain, etc.

I see. thanks, bro.