search

csxmli2016 / DFDNet

Blind Face Restoration via Deep Multi-scale Component Dictionaries (ECCV 2020)
915 stars 215 forks source link

Request (Only Face Restoration) #41

Closed blanuk closed 4 years ago

blanuk commented 4 years ago

Hi :)

I tried it out and and the result is awesome :) But: i need only step 3 and 4. Means no aligment, no landmark.

How can i do that?

Thanks a lot :)

//edit: the source comes already aligned and cropped to 512x512 :) (DFL from iperov)

csxmli2016 commented 4 years ago

)

Hi, landmark is essential, we need it to crop the eyes, nose and mouth region for dictionary feature transformation. As for alignment, I think you'd better use our method. Because the training data is aligned in a fixed location, other alignments will generate unpleasant results due to the inconsistent features with training data.

blanuk commented 4 years ago

okay, i see. thanks a lot 👍 well, i have to take a deeper look how to use it in my workflow :)