Skip to content


Switch branches/tags

Name already in use

A tag already exists with the provided branch name. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. Are you sure you want to create this branch?


Failed to load latest commit information.
Latest commit message
Commit time
Sep 21, 2020
Sep 21, 2020
Dec 6, 2020
Dec 14, 2021
Dec 6, 2020
Dec 6, 2020
Sep 21, 2020

[ECCV 2020] Flow-edge Guided Video Completion

[Paper] [Project Website] [Google Colab]

We present a new flow-based video completion algorithm. Previous flow completion methods are often unable to retain the sharpness of motion boundaries. Our method first extracts and completes motion edges, and then uses them to guide piecewise-smooth flow completion with sharp edges. Existing methods propagate colors among local flow connections between adjacent frames. However, not all missing regions in a video can be reached in this way because the motion boundaries form impenetrable barriers. Our method alleviates this problem by introducing non-local flow connections to temporally distant frames, enabling propagating video content over motion boundaries. We validate our approach on the DAVIS dataset. Both visual and quantitative results show that our method compares favorably against the state-of-the-art algorithms.

[ECCV 2020] Flow-edge Guided Video Completion
Chen Gao, Ayush Saraf, Jia-Bin Huang, and Johannes Kopf
In European Conference on Computer Vision (ECCV), 2020


  • Linux (tested on CentOS Linux release 7.4.1708)
  • Anaconda
  • Python 3.8 (tested on 3.8.5)
  • PyTorch 1.6.0

and the Python dependencies listed in requirements.txt

  • To get started, please run the following commands:

    conda create -n FGVC
    conda activate FGVC
    conda install pytorch=1.6.0 torchvision=0.7.0 cudatoolkit=10.1 matplotlib scipy opencv -c pytorch
    pip install imageio imageio-ffmpeg scikit-image imutils
  • Next, please download the model weight and demo data using the following command:

    chmod +x

Quick start

  • Object removal:
cd tool
python \
       --mode object_removal \
       --path ../data/tennis \
       --path_mask ../data/tennis_mask \
       --outroot ../result/tennis_removal \
  • FOV extrapolation:
cd tool
python \
       --mode video_extrapolation \
       --path ../data/tennis \
       --outroot ../result/tennis_extrapolation \
       --H_scale 2 \
       --W_scale 2 \

You can remove the --seamless flag for a faster processing time.


This work is licensed under MIT License. See LICENSE for details.

If you find this code useful for your research, please consider citing the following paper:

    author    = {Gao, Chen and Saraf, Ayush and Huang, Jia-Bin and Kopf, Johannes},
    title     = {Flow-edge Guided Video Completion},
    booktitle = {European Conference on Computer Vision},
    year      = {2020}


  • Our flow edge completion network builds upon EdgeConnect.
  • Our image inpainting network is modified from DFVI.


[ECCV 2020] Flow-edge Guided Video Completion







No releases published


No packages published