NeRViS: Neural Re-rendering for Full-frame Video Stabilization

Project Page | Video | Paper | Google Colab

Setup

Setup environment for [Yu and Ramamoorthi 2020].

cd CVPR2020CODE_yulunliu_modified
conda create --name NeRViS_CVPR2020 python=3.6
conda activate NeRViS_CVPR2020
pip install -r requirements_CVPR2020.txt
./install.sh

Download pre-trained checkpoints of [Yu and Ramamoorthi 2020].

wget https://www.cmlab.csie.ntu.edu.tw/~yulunliu/NeRViS/CVPR2020_ckpts.zip
unzip CVPR2020_ckpts.zip
cd ..

Setup environment for NeRViS.

conda deactivate
conda create --name NeRViS python=3.6
conda activate NeRViS
conda install pytorch=1.6.0 torchvision=0.7.0 cudatoolkit=10.1 -c pytorch
conda install matplotlib
conda install tensorboard
conda install scipy
conda install opencv
conda install -c conda-forge cupy cudatoolkit=10.1
pip install PyMaxflow

Running code

Calculate smoothed flow using [Yu and Ramamoorthi 2020].

conda activate NeRViS_CVPR2020
cd CVPR2020CODE_yulunliu_modified
python main.py [input_frames_path] [output_frames_path] [output_warping_field_path]

e.g.

python main.py ../../NUS/Crowd/0/ NUS_results/Crowd/0/ CVPR2020_warping_field/

Run NeRViS video stabilization.

conda deactivate
conda activate NeRViS
cd ..
python run_NeRViS.py --load [model_checkpoint_path] --input_frames_path [input_frames_path] --warping_field_path [warping_field_path] --output_path [output_frames_path] --temporal_width [temporal_width] --temporal_step [temporal_step]

e.g.

python run_NeRViS.py --load NeRViS_model/checkpoint/model_epoch050.pth --input_frames_path ../NUS/Crowd/0/ --warping_field_path CVPR2020CODE_yulunliu_modified/CVPR2020_warping_field/ --output_path output/ --temporal_width 41 --temporal_step 4

Citation

@inproceedings{Liu-NeRViS-2021,
    author    = {Liu, Yu-Lun and Lai, Wei-Sheng and Yang, Ming-Hsuan and Chuang, Yung-Yu and Huang, Jia-Bin}, 
    title     = {Neural Re-rendering for Full-frame Video Stabilization}, 
    journal   = {arXiv preprint},
    year      = {2021}
}

Acknowledgements

Parts of the code were based on from AdaCoF-pytorch. Some functions are borrowed from softmax-splatting, RAFT, and [Yu and Ramamoorthi 2020]