Easy use of good super resolution algorithms, now only EDVR from BasicSR can be used, No mater how big the input is, it can process, so I'm going to release it now even it's still not compelete.
Colab Demo: Notebooks
- NVIDIA GPU + CUDA (We test with: V100, P100, P4, T4, K80)
- Ubuntu (We test with Ubuntu = 18.04.5 LTS)
- FFmpeg
- CUDA (We test with CUDA = 10.1)
- cuDNN (Optional)
- Python >= 3.7 (Recommend to use Anaconda or Miniconda)
- PyTorch >= 1.3
- Install VSR
git clone https://github.com/iBobbyTS/VSR.git cd VSR python build.py
- Download models
cd BasicSR python scripts/download_pretrained_models.py EDVR cd ..
```
cd VSR
python run.py -i input.mp4
```
Check Other arguements in run.py.
See MIT License