Combination of DAIN, Super SloMo(SSM), BIN and more coming together. Now SSM and DAIN are developed, BIN is still being worked on.

Colab Demo: Notebooks Built VFIN

1. Citation
2. Requirements and Dependencies
3. Installation
4. Inferencing

Frame interpolation for normal video @inproceedings{DAIN, author = {Bao, Wenbo and Lai, Wei-Sheng and Ma, Chao and Zhang, Xiaoyun and Gao, Zhiyong and Yang, Ming-Hsuan}, title = {Depth-Aware Video Frame Interpolation}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2019} }

Frame interpolation for general video @inproceedings{Super SloMo, author = {Jiang, Huaizu and Sun, Deqing and Jampani, Varun and Yang, Ming-Hsuan and Learned-Miller, Erik and Kautz, Jan}, title = {High Quality Estimation of Multiple Intermediate Frames for Video Interpolation}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2018} }

Frame interpolation for blurry video @inproceedings{BIN, author = {Shen, Wang and Bao, Wenbo and Zhai, Guangtao and Chen, Li and Min, Xiongkuo and Gao, Zhiyong}, title = {Blurry Video Frame Interpolation}, booktitle = {IEEE Conference on Computer Vision and Pattern Recognition}, year = {2020} }

• General
  • FFmpeg
  • NVIDIA GPU (We test with: P100, P4, T4, K80)
• DAIN
  • Ubuntu (We test with Ubuntu = 18.04.5 LTS)
  • Python >= 3.6, <= 3.8
  • CUDA and cuDNN (We test with CUDA = 10.1)
  • PyTorch >= 1.0.0, <= 1.4.0
  • GCC (Compiling PyTorch 1.0.0 extension files (.c/.cu) requires gcc = 4.9.1 and nvcc = 9.0 compilers)
• SSM
  • Ubuntu (We test with Ubuntu = 18.04.5 LTS)
  • Python (We test with Python = 3.6.12)
  • PyTorch >= 0.4.1, <= 1.7.0

Install dependencies:

# Python
pip install numpy opencv-python
# Change the cuda version to your version. ex. CUDA 9.2: +cu92, or CPU, +cpu
pip install torch==1.4.0+cu101 torchvision==0.5.0+cu101 -f https://download.pytorch.org/whl/torch_stable.html

# Conda (Anaconda/Miniconda)
conda install numpy -y
pip install opencv-python
# Change cudatoolkit to your version
conda install pytorch==1.4.0 torchvision==0.5.0 cudatoolkit=10.1 -c pytorch -y

Download repository:

git clone https://github.com/iBobbyTS/VFIN.git
cd VFIN

Generate our PyTorch extensions: (This will take approximately 20 minuets) Check build.py for args available.

cd DAIN
python build.py
cd ..

Download pretrained models,

# DAIN
mkdir DAIN/model_weights
wget http://vllab1.ucmerced.edu/~wenbobao/DAIN/best.pth -O DAIN/model_weights/best.pth
# SSM (wget might not work, find other wayto download it from Google Drive and copy it to SSM)
wget https://drive.google.com/file/d/1IvobLDbRiBgZr3ryCRrWL8xDbMZ-KnpF/view?usp=drive_open -O SSM/SuperSloMo.ckpt

python run.py -i input.mp4

Check Other arguements in run.py.

iBobby

See MIT License