DAN-Basd-on-Openmmlab

DAN: Unfolding the Alternating Optimization for Blind Super Resolution

We reproduce DAN via mmediting based on open-sourced code.

Requirements

• PyTorch >= 1.3
• mmediting >= 0.9

DataSets

We use DIV2K and Flickr2K as our training datasets. For evaluation of Setting 2, we use DIV2KRK datasets,

Usages

How to run this repo: copy the file to the mmediting workspace and run the program directly based on the commands in mmediting

1. Copy files to MMEditing workspace.

cd DAN-Basd-on-Openmmlab/
mv ./mmedit/models/restorers/dan.py ${mmediting_workspace}/mmedit/models/restorers/
mv ./mmedit/models/backbones/sr_backbones/dan_net.py ${mmediting_workspace}/mmedit/models/backbones/sr_backbones/
mv ./mmedit/models/common/DANpreprocess.py ${mmediting_workspace}/mmedit/models/common
mv ./configs/restorers/dan ${mmediting_workspace}/configs/restorers/
mv ./tools/data/super-resolution/dan_datasets ${mmediting_workspace}/tools/data/super-resolution/

2. Modify the configuration file as follows:

pca_matrix_path='${mmediting_workspace}/tools/data/super-resolution/div2k/pca_matrix/pca_aniso_matrix_x4.pth' # your pca_matrix path
# Training
gt_folder='${dataset_workspace}/dataset/DF2K_train_HR_sub' # your train data path
# Testing
lq_folder='${dataset_workspace}/dataset/DIV2KRK/lr_x4' # your test data LR path
gt_folder='${dataset_workspace}/dataset/DIV2KRK/gt' # your test data HR path

3. Add script to init file, as follows:

• modify the mmedit/models/backbones/sr_backbones/__init__.py:

from .dan_net import DAN
# add DAN into __all__ list.

• modify the mmedit/models/commons/__init__.py:

from .dan_preprocess import SRMDPreprocessing
# add SRMDreprocessing into __all__ list.

• modify the mmedit/models/restorers/__init__.py:

from .dan import DAN
# add DAN into __all__ list.

4. Training/Test

Before using it, please download and process the dataset and set the path in the configuration file.

• Train

# Single GPU
python tools/train.py configs/restorers/dan/dan_setting2.py --work_dir ${YOUR_WORK_DIR}

# Multiple GPUs
./tools/dist_train.sh configs/restorers/dan/dan_setting2.py ${GPU_NUM} --work_dir ${YOUR_WORK_DIR}

• Test

# Single GPU
python tools/test.py configs/restorers/dan/dan_setting2.py ${CHECKPOINT_FILE} [--metrics ${METRICS}] [--out ${RESULT_FILE}]

# Multiple GPUs
./tools/dist_test.sh configs/restorers/dan/dan_setting2.py ${CHECKPOINT_FILE} ${GPU_NUM} [--metrics ${METRICS}] [--out ${RESULT_FILE}]

Result

DIV2KRK

The passwds of download links are all 'ta2o'

Method	scale	Datasets	PSNR	SSIM	Download
DAN-RGB (paper)	x4	DIV2KRK	26.09	0.7312	-
DAN-Y (paper)	x4	DIV2KRK	27.55	0.7582	-
DAN-RGB (Ours)	x4	DIV2KRK	27.41	0.7666	model / test_pkl
DAN-Y (Ours)	x4	DIV2KRK	28.88	0.7915	model / test_pkl

---