search

wyhuai / DDNM

[ICLR 2023 Oral] Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model
MIT License
1.16k stars 83 forks source link

Super resolution quaility on arbitrary imagenet image #37

Open zkaiWu opened 1 year ago

zkaiWu commented 1 year ago

I test ddnm on arbitrary imagenet image but did not get as good results as the demo. The command I use is : CUDA_VISIBLE_DEVICES=3 python main.py --resize_y --config confs/inet256.yml --path_y ./ILSVRC2012_val_00046632.JPEG --class 22 --deg "sr_averagepooling" --scale 4 -i imagenet_46632_cc_256x96

Apy: 00000_apy

final: 00000

Is there any parameter I need to adjust?

wyhuai commented 1 year ago

Hi, this is because of the out-of-domain problem. For example, the first patch start from the left-up corner, where no eagles appear. Actually, none of a patch can cover full eagle content. In this case, using Hierarchical Restoration may alleviate this problem. You may refer to the paper for Hierarchical Restoration Unlimited-Size Diffusion Restoration.