search

wyhuai / DDNM

[ICLR 2023 Oral] Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model
MIT License
1.15k stars 82 forks source link
diffusion-models iclr iclr2023 image-restoration zero-shot

readme

Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model

📖Paper|🖼️Project Page| google colab logo Hugging Face Spaces

Yinhuai Wang*, Jiwen Yu*, Jian Zhang
Peking University and PCL
*denotes equal contribution

This repository contains the code release for *Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model*. DDNM can solve various image restoration tasks without any optimization or training! Yes, in a zero-shot manner.

Supported Applications:

front

🧩News

Installation

Code

git clone https://github.com/wyhuai/DDNM.git

Environment

pip install numpy torch blobfile tqdm pyYaml pillow    # e.g. torch 1.7.1+cu110.

Pre-Trained Models

To restore human face images, download this model(from SDEdit) and put it into DDNM/exp/logs/celeba/. 

https://drive.google.com/file/d/1wSoA5fm_d6JBZk4RZ1SzWLMgev4WqH21/view?usp=share_link

To restore general images, download this model(from guided-diffusion) and put it into DDNM/exp/logs/imagenet/.

wget https://openaipublic.blob.core.windows.net/diffusion/jul-2021/256x256_diffusion_uncond.pt

Quick Start

Run below command to get 4x SR results immediately. The results should be in DDNM/exp/image_samples/demo.

python main.py --ni --simplified --config celeba_hq.yml --path_y celeba_hq --eta 0.85 --deg "sr_averagepooling" --deg_scale 4.0 --sigma_y 0 -i demo

Setting

The detailed sampling command is here:

python main.py --ni --simplified --config {CONFIG}.yml --path_y {PATH_Y} --eta {ETA} --deg {DEGRADATION} --deg_scale {DEGRADATION_SCALE} --sigma_y {SIGMA_Y} -i {IMAGE_FOLDER}

with following options:

For the config files, e.g., celeba_hq.yml, you may change following properties:

sampling:
    batch_size: 1

time_travel:
    T_sampling: 100     # sampling steps
    travel_length: 1    # time-travel parameters l and s, see section 3.3 of the paper.
    travel_repeat: 1    # time-travel parameter r, see section 3.3 of the paper.

Reproduce The Results In The Paper

Quantitative Evaluation

Dataset download link: [Google drive] [Baidu drive]

Download the CelebA testset and put it into DDNM/exp/datasets/celeba/.

Download the ImageNet testset and put it into DDNM/exp/datasets/imagenet/ and replace the file DDNM/exp/imagenet_val_1k.txt.

Run the following command. You may increase the batch_size to accelerate evaluation.

sh evaluation.sh

High-Quality Results

You can try this Colab demo for High-Quality results. Note that the High-Quality results presented in the front figure are mostly generated by applying DDNM to the models in RePaint.

🔥Real-World Applications

Demo: Real-World Super-Resolution.

orig_62 00000

Run the following command

python main.py --ni --simplified --config celeba_hq.yml --path_y solvay --eta 0.85 --deg "sr_averagepooling" --deg_scale 4.0 --sigma_y 0.1 -i demo

Demo: Old Photo Restoration.

image image

Run the following command

python main.py --ni --simplified --config oldphoto.yml --path_y oldphoto --eta 0.85 --deg "mask_color_sr" --deg_scale 2.0 --sigma_y 0.02 -i demo

Try your own photos.

You may use DDNM to restore your own degraded images. DDNM provides full flexibility for you to define the degradation operator and the noise level. Note that these definitions are critical for a good results. You may reference the following guidance.

  1. If you are using CelebA pretrained models, try the tool in GFPGAN to crop and align your photo.
  2. If there are local artifacts on your photo, try the tool in LaMa to draw a mask to cover them, and save this mask to DDNM/exp/inp_masks/mask.png. Then run DDNM/exp/inp_masks/get_mask.py to generate mask.npy.
  3. If your photo is faded, you need a grayscale operator as part of the degradation.
  4. If your photo is blur, you need a downsampler operator as part of the degradation. Also, you need to set a proper SR scale --deg_scale.
  5. If your photo suffers global artifacts, e.g., jpeg-like artifacts or unkown noise, you need to set a proper sigma_y to remove these artifacts.
  6. Search args.deg =='diy' in DDNM/guided_diffusion/diffusion.py and change the definition of $\mathbf{A}$ correspondingly. Then run 
    python main.py --ni --simplified --config celeba_hq.yml --path_y {YOUR_OWN_PATH} --eta 0.85 --deg "diy" --deg_scale {YOUR_OWN_SCALE} --sigma_y {YOUR_OWN_LEVEL} -i diy

🆕DDNM for Arbitrary Size

123456image

Above we show an example of using DDNM to SR a 64x256 input image into a 256x1024 result. The theory details can be found in this paper, section 3.3.

We implement the Mask-Shift Restoration in the folder hq_demo, based on RePaint. You can try this Colab demo. google colab logo

Or, you can try this function on your own device, you need to download the pre-trained models:

wget https://openaipublic.blob.core.windows.net/diffusion/jul-2021/256x256_classifier.pt
wget https://openaipublic.blob.core.windows.net/diffusion/jul-2021/256x256_diffusion.pt

and put it to hq_demo/data/pretrained. Then run

cd hq_demo
sh evaluation.sh

This script contains SR results up to 2K resolution. It may take hours to finish some demos in this script. Setting a smaller sampling step or time-travel parameters in hq_demo/confs/inet256.yml can speed up, but may compromise the generative quality.

😊Applying DDNM to Your Own Diffusion Model

It is very easy to implement a basic DDNM on your own diffusion model! You may reference the following:

  1. Copy these operator implementations to the core diffusion sampling file, then define your task type, e.g., set IR_mode="super resolution". 
    
def color2gray(x):
coef=1/3
x = x[:,0,:,:] * coef + x[:,1,:,:]*coef +  x[:,2,:,:]*coef
return x.repeat(1,3,1,1)

def gray2color(x): x = x[:,0,:,:] coef=1/3 base = coef2 + coef2 + coef*2 return th.stack((xcoef/base, xcoef/base, xcoef/base), 1)

def PatchUpsample(x, scale): n, c, h, w = x.shape x = torch.zeros(n,c,h,scale,w,scale) + x.view(n,c,h,1,w,1) return x.view(n,c,scaleh,scalew)

Implementation of A and its pseudo-inverse Ap

if IR_mode=="colorization": A = color2gray Ap = gray2color

elif IR_mode=="inpainting": A = lambda z: z*mask Ap = A

elif IR_mode=="super resolution": A = torch.nn.AdaptiveAvgPool2d((256//scale,256//scale)) Ap = lambda z: PatchUpsample(z, scale)

elif IR_mode=="old photo restoration": A1 = lambda z: z*mask A1p = A1

A2 = color2gray
A2p = gray2color

A3 = torch.nn.AdaptiveAvgPool2d((256//scale,256//scale))
A3p = lambda z: PatchUpsample(z, scale)

A = lambda z: A3(A2(A1(z)))
Ap = lambda z: A1p(A2p(A3p(z)))
2. Find the variant $\mathbf{x}\_{0|t}$ in the target codes, and use the result of this function to modify the sampling of $\mathbf{x}\_{t-1}$. Your may need to provide the input degraded image $\mathbf{y}$ and the corresponding noise level $\sigma_\mathbf{y}$.
```python
# Core Implementation of DDNM+, simplified denoising solution (Section 3.3).
# For more accurate denoising, please refer to the paper (Appendix I) and the source code.

def ddnm_plus_core(x0t, y, sigma_y=0, sigma_t, a_t):

    #Eq 19
    if sigma_t >= a_t*sigma_y: 
        lambda_t = 1
        gamma_t = sigma_t**2 - (a_t*lambda_t*sigma_y)**2
    else:
        lambda_t = sigma_t/(a_t*sigma_y)
        gamma_t = 0

    #Eq 17    
    x0t= x0t + lambda_t*Ap(y - A(x0t))

    return x0t, gamma_t
  1. Actually, this repository contains the above simplified implementation: try search arg.simplified in DDNM/guided_diffusion/diffusion.py for related codes.

References

If you find this repository useful for your research, please cite the following work.

@article{wang2022zero,
  title={Zero-Shot Image Restoration Using Denoising Diffusion Null-Space Model},
  author={Wang, Yinhuai and Yu, Jiwen and Zhang, Jian},
  journal={The Eleventh International Conference on Learning Representations},
  year={2023}
}

This implementation is based on / inspired by: