This repository is for the paper "Improving Bi-direction Recurrent Network for Video Super-Resolution with Deformable Motion Alignment Structure "
- Please refer to DatasetPreparation.md for more details.
- The descriptions of currently supported datasets (
torch.utils.data.Dataset
classes) are in Datasets.md.
- Python >= 3.7 (Recommend to use Anaconda or Miniconda)
- PyTorch >= 1.7
- NVIDIA GPU + CUDA
-
Clone repository
git clone https://github.com/hengliusky/IBRN_VSR.git
-
Install dependent packages
cd IBRN_VSR pip install -r requirements.txt
-
Compile basicsr environment
python setup.py develop
# Train on REDS
sh train.sh reds
# train on vimeo-90k
sh train.sh vimeo90k
# train on RealVSR
sh train.sh real
# test on reds
sh test.sh reds
# test on vimeo
sh test.sh vimeo90k
# test on vid4
sh test.sh vid4
# test on RealVSR
sh test.sh real
The visualization clustering comparisons on the Cora, Citeseer and Pubmed of our proposed approach with R-DGAE, MAGCN and GMM-VGAE.