hengliusky/IBRN_VSR

Title

This repository is for the paper "Improving Bi-direction Recurrent Network for Video Super-Resolution with Deformable Motion Alignment Structure "

Dataset Preparation

• Please refer to DatasetPreparation.md for more details.
• The descriptions of currently supported datasets (torch.utils.data.Dataset classes) are in Datasets.md.

Dependencies and Installation

• Python >= 3.7 (Recommend to use Anaconda or Miniconda)
• PyTorch >= 1.7
• NVIDIA GPU + CUDA

1. Clone repository

   git clone https://github.com/hengliusky/IBRN_VSR.git

2. Install dependent packages

   cd IBRN_VSR
   pip install -r requirements.txt

3. Compile basicsr environment

   python setup.py develop

Training

# Train on REDS
sh train.sh reds

# train on vimeo-90k
sh train.sh vimeo90k

# train on RealVSR
sh train.sh real

Testing

# 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

Results

The visualization clustering comparisons on the Cora, Citeseer and Pubmed of our proposed approach with R-DGAE, MAGCN and GMM-VGAE.

Vid4

Calendar

Foliage

REDS

Clip 000

Clip 015

Temporal Consistent