UHDDIP: Ultra-High-Definition Restoration: New Benchmarks and A Dual Interaction Prior-Driven Solution
| Download Link | Description |
|---|---|
| Google Drive (https://drive.google.com/drive/u/1/folders/1LaQvEBdjH5MwTwkfCZh3UJUGl8mYfvke?hl=zh_CN)) | A total of 3000 pairs for training and 200 pairs for testing. |
- CUDA 10.1 (or later)
- Python 3.9 (or later)
- Pytorch 1.8.1 (or later)
- Torchvision 0.19
- OpenCV 4.7.0
- tensorboard, skimage, scipy, lmdb, tqdm, yaml, einops, natsort
- Please download the following datasets:
| Task | Training dataset | Testing dataset |
|---|---|---|
| UHD LLIE | UHD_LL (2000) | UHD_LL (150) |
| UHD Desnowing | UHD_Snow (3000) | UHD_Snow (200) |
| UHD Deraining | UHD_Rain (3000) | UHD_Rain (200) |
- To obtain the normal prior images, all training data and testing data are processed through Omnidata, and the obtained normal prior images are placed under the same path as the corresponding input image and GT image above.
-
To train UHD IILE model, modify the path where the UHD_LL dataset (including input, gt, normal) is located in the /src/Options/train_UHDDIP.yml, then run
cd UHDDIP python -m torch.distributed.launch --nproc_per_node=2 --master_port=4321 basicsr/train.py -opt src/Options/train_UHDDIP.yml --launcher pytorch -
To train UHD desnowing model, modify the path where the UHD_Snow dataset (including input, gt, normal) is located in the /src/Options/train_UHDDIP.yml, then run
cd UHDDIP python -m torch.distributed.launch --nproc_per_node=2 --master_port=4321 basicsr/train.py -opt src/Options/train_UHDDIP.yml --launcher pytorch -
To train UHD deraining model, modify the path where the UHD_Rain dataset (including input, gt, normal) is located in the /src/Options/train_UHDDIP.yml, then run
cd UHDDIP python -m torch.distributed.launch --nproc_per_node=2 --master_port=4321 basicsr/train.py -opt src/Options/train_UHDDIP.yml --launcher pytorch
-
Download all pre-trained models and place them in
./pretrained_models/ -
Testing
cd src python test_uhd.py -
Calculating PSNR/SSIM/LPIPS scores, run
python calculate_psnr_ssim.py
| Task | Model | Visual Results |
|---|---|---|
| UHD LLIE | Download | Download |
| UHD Desnowing | Download | Download |
| UHD Deraining | Download | Download |
UHD Image LLIE (click to expand)
UHD Image Desnowing (click to expand)
UHD Image Deraining (click to expand)
