junjun-jiang / face-hallucination-benchmark Goto Github PK

I think it should be GFP-GAN rather than CFP-GAN.

• [CFP-GAN] Towards Real-World Blind Face Restoration with Generative Facial Prior, X. Wang et al., CVPR 2021.

Hello!
Could you please provide the training sets of celeba? The picture I cut is inconsistent with the picture of the test sets you provided.
Thank you very much!

Hi,

Thank you for your wonderful work. It is really helpful and saved me a lot of time!

I tried to test the produced SR images (from some SOTA methods) you provided on Baidu Driver, and computed PSNR and SSIM using the code in this implementation (eval_psnr_ssim.py). However, what I obtained were quiet different from the results reported in your survey paper (Table 7). For example, the RCAN method achieved about 31.50dB on CelebA x4, but the result in Table 7 is 31.69dB.

I was wondering if I missed something important. Thank you, and look forward to ur reply : )

Is there anything new in 2022?
Looking forward to SOTA works in 2022 .
Thanks :)

Thank you so much for this good job! This wonderful job has saved me a lot time to find papers about face hallucination. But is there any new papers in 2019? (I haven't found any ...)