DiffBIR: A Python repository from treksis

DiffBIR: Towards Blind Image Restoration with Generative Diffusion Prior

Paper | Project Page

Xinqi Lin^1,*, Jingwen He^2,*, Ziyan Chen², Zhaoyang Lyu², Ben Fei², Bo Dai², Wanli Ouyang², Yu Qiao², Chao Dong^1,2

¹Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences
²Shanghai AI Laboratory

⭐If DiffBIR is helpful for you, please help star this repo. Thanks!🤗

👀Visual Results On Real-world Images

General Image Restoration

Face Image Restoration

⚙️Installation

Python >= 3.9
CUDA >= 11.3
PyTorch >= 1.12.1
xformers == 0.0.16

# clone this repo
git clone https://github.com/XPixelGroup/DiffBIR.git
cd DiffBIR

# create a conda environment with python >= 3.9
conda create -n diffbir python=3.9
conda activate diffbir

conda install pytorch==1.12.1 torchvision==0.13.1 cudatoolkit=11.3 -c pytorch
conda install xformers==0.0.16 -c xformers

# other dependencies
pip install -r requirements.txt

🧬Pretrained Models

Model Name	Description	HuggingFace	BaiduNetdisk
general_swinir_v1.ckpt	Stage1 model (SwinIR) for general image restoration.	download	download (pwd: v3v6)
general_full_v1.ckpt	Full model for general image restoration. "Full" means it contains both the stage1 and stage2 model.	download	download (pwd: 86zi)
face_swinir_v1.ckpt	Stage1 model (SwinIR) for face restoration.	download	download (pwd: xk5u)
face_full_v1.ckpt	Full model for face restoration.	download	download (pwd: ov8i)

🛫Quick Start

Download general_full_v1.ckpt and general_swinir_v1.ckpt to weights/, then run the following command to interact with the gradio website.

python gradio_diffbir.py \
--ckpt weights/general_full_v1.ckpt \
--config configs/model/cldm.yaml \
--reload_swinir \
--swinir_ckpt weights/general_swinir_v1.ckpt \
--device cuda

⚔️Inference

Full Pipeline (Remove Degradations & Refine Details)

General Image

Download general_full_v1.ckpt and general_swinir_v1.ckpt to weights/ and run the following command.

python inference.py \
--input inputs/general \
--config configs/model/cldm.yaml \
--ckpt weights/general_full_v1.ckpt \
--reload_swinir --swinir_ckpt weights/general_swinir_v1.ckpt \
--steps 50 \
--sr_scale 4 \
--image_size 512 \
--color_fix_type wavelet --resize_back \
--output results/general \
--device cuda

If you are confused about where the reload_swinir option came from, please refer to the degradation details.

Face Image

Download face_full_v1.ckpt to weights/ and run the following command.

# for aligned face inputs
python inference_face.py \
--config configs/model/cldm.yaml \
--ckpt weights/face_full_v1.ckpt \
--input inputs/face/aligned \
--steps 50 \
--sr_scale 1 \
--image_size 512 \
--color_fix_type wavelet \
--output results/face/aligned --resize_back \
--has_aligned \
--device cuda

# for unaligned face inputs
python inference_face.py \
--config configs/model/cldm.yaml \
--ckpt weights/face_full_v1.ckpt \
--input inputs/face/whole_img \
--steps 50 \
--sr_scale 1 \
--image_size 512 \
--color_fix_type wavelet \
--output results/face/whole_img --resize_back \
--device cuda

Only Stage1 Model (Remove Degradations)

Download general_swinir_v1.ckpt, face_swinir_v1.ckpt for general, face image respectively, and run the following command.

python scripts/inference_stage1.py \
--config configs/model/swinir.yaml \
--ckpt [swinir_ckpt_path] \
--input [lq_dir] \
--sr_scale 1 --image_size 512 \
--output [output_dir_path]

Only Stage2 Model (Refine Details)

Since the proposed two-stage pipeline is very flexible, you can utilize other awesome models to remove degradations instead of SwinIR and then leverage the Stable Diffusion to refine details.

# step1: Use other models to remove degradations and save results in [img_dir_path].

# step2: Refine details of step1 outputs.
python inference.py \
--config configs/model/cldm.yaml \
--ckpt [full_ckpt_path] \
--steps 50 --sr_scale 1 --image_size 512 \
--input [img_dir_path] \
--color_fix_type wavelet --resize_back \
--output [output_dir_path] \
--disable_preprocess_model \
--device cuda

🌠Train

Degradation Details

For general image restoration, we first train both the stage1 and stage2 model under codeformer degradation to enhance the generative capacity of the stage2 model. In order to improve the ability for degradation removal, we train another stage1 model under Real-ESRGAN degradation and utilize it during inference.

For face image restoration, we adopt the degradation model used in DifFace for training and directly utilize the SwinIR model released by them as our stage1 model.

Data Preparation

Generate file list of training set and validation set.
```
python scripts/make_file_list.py \
--img_folder [hq_dir_path] \
--val_size [validation_set_size] \
--save_folder [save_dir_path] \
--follow_links
```
This script will collect all image files in img_folder and split them into training set and validation set automatically. You will get two file lists in save_folder, each line in a file list contains an absolute path of an image file:
```
save_folder
├── train.list # training file list
└── val.list   # validation file list
```
Configure training set and validation set.

For general image restoration, fill in the following configuration files with appropriate values.
- training set and validation set for CodeFormer degradation.
- training set and validation set for Real-ESRGAN degradation.
For face image restoration, fill in the face training set and validation set configuration files with appropriate values.

Train Stage1 Model

Configure training-related information.

Fill in the configuration file of training with appropriate values.
Start training.
```
python train.py --config [training_config_path]
```
💡:Checkpoints of SwinIR will be used in training stage2 model.

Train Stage2 Model

Download pretrained Stable Diffusion v2.1 to provide generative capabilities.

wget https://huggingface.co/stabilityai/stable-diffusion-2-1-base/resolve/main/v2-1_512-ema-pruned.ckpt --no-check-certificate

Create the initial model weights.

python scripts/make_stage2_init_weight.py \
--cldm_config configs/model/cldm.yaml \
--sd_weight [sd_v2.1_ckpt_path] \
--swinir_weight [swinir_ckpt_path] \
--output [init_weight_output_path]

You will see some outputs which show the weight initialization.

Configure training-related information.

Fill in the configuration file of training with appropriate values.

Start training.

python train.py --config [training_config_path]

🆕Update

2023.08.30: Repo is released.
2023.09.06: Update colab demo. Thanks to camenduru!:hugs:
2023.09.08: Add support for restoring unaligned faces.

🧗TODO

Release code and pretrained models:computer:.
Update links to paper and project page:link:.
Release real47 testset:minidisc:.
Reduce the memory usage of DiffBIR:smiley_cat:.
Provide HuggingFace demo:notebook:.
Upload inference code of latent image guidance:page_facing_up:.
Improve the performance:superhero:.
Add a patch-based sampling schedule:mag:.

Citation

Please cite us if our work is useful for your research.

@article{2023diffbir,
  author    = {Xinqi Lin, Jingwen He, Ziyan Chen, Zhaoyang Lyu, Ben Fei, Bo Dai, Wanli Ouyang, Yu Qiao, Chao Dong},
  title     = {DiffBIR: Towards Blind Image Restoration with Generative Diffusion Prior},
  journal   = {arxiv},
  year      = {2023},
}

License

This project is released under the Apache 2.0 license.

Acknowledgement

This project is based on ControlNet and BasicSR. Thanks for their awesome work.

Contact

If you have any questions, please feel free to contact with me at linxinqi@tju.edu.cn.

treksis/DiffBIR