Dependencies and Installation
- The code is based on BasicSR, Please install the BasicSR framework first.
- Pytorch=1.51
Training
cd ./code
CUDA_VISIBLE_DEVICES=0,1 python basicsr/train.py -opt options/train/BasicVSR/train_BasicVSR.yml
CUDA_VISIBLE_DEVICES=0,1 python basicsr/train.py -opt options/train/BasicVSR/train_IconVSR.yml
Testing
cd ./code
CUDA_VISIBLE_DEVICES=0 python basicsr/test.py -opt options/test/BasicVSR/test_BasicVSR_REDS.yml
CUDA_VISIBLE_DEVICES=0 python basicsr/test.py -opt options/test/BasicVSR/test_BasicVSR_Vid4.yml
PSNR/SSIM Results
It takes about 5 days to train the BasicVSR/IconVSR model with the REDS dataset on 2 V100 GPUs.
Dataset(BI) | BasicVSR (paper) | Ours | IconVSR_w/o Refill (paper) | Ours |
---|---|---|---|---|
REDS4 | 31.42/0.8909 | 31.409/0.8907 | 31.60 | 31.6026 |
Vid4 | 27.24/0.8251 | 27.269/0.8311 | - | - |
- Pretrained models and SR results can be downloaded Here.