ir1d / afn Goto Github PK

I wanted to test the model quickly and wasn't confident in the level of my training model. Could you share your model

AFN paper states the following:

• "trackAFN (Ours) 34.20 0.9392" (Table 5: Quantitative results on NTIRE20 Challenge Deblur)
• "Our AFN has 47.7M parameters, and requires 760G MACs for an input image of size 128 × 128" (4.2. Implementation Details)

AFN competed in track1 and track2 of "NTIRE 2020 Challenge on Image and Video Deblurring" with the results:

• track1: "OIerM Attentive Fractal Network 34.20 0.9392 1.16" (Table 1: Single image deblurring results on the REDS test data)
• track2: "OIerM Attentive Fractal Network 28.33 0.8079 0.9" (Table 2: Single image deblurring results on the REDS test data from Google Pixel 4)

I have 2 questions:

1. what is the diff between AFN network from track1 and AFN network from track2?
2. where can I download the CFAMoire dataset?

Thank you!

hi， it's a pleasure reading your paper, In your readme, I got the website of NTIRE2020 dataset , But I can't downloading the dataset(Maybe it's because the competition is over )， Could you share me the dataset, thanx

Hello.
Could you share the pretrained model, please?

Thanks for your great work for Demoireing. After reading your paper and the code. I still have some question. In the paper, is AFN++ a two cascaded AFN1(stage 1) and AFN2(stage 2) network? When the AFN1 trained completely, and then start training the AFN2. Does this mean the parameters of AFN1 in the stage 2 are fixed and are’t updated? Only upade the AFN2’s parameters?
Also, in your paper, in the stage 2, “augment the input image using flip, transpose and rotate operations” and get the responding generating coarse results from AFN1. But I haven’t seen the related code for the stage 2. Do I miss something else?
Thanks for your reply.