Package containing helper functions for loading and evaluating DAVIS
Switch branches/tags
Nothing to show
Clone or download
Pull request Compare This branch is 1 commit ahead of fperazzi:master.
Latest commit a44a351 May 15, 2018
Type Name Latest commit message Commit time
Failed to load latest commit information.
cmake Replicate matlab implementation of t_stability Apr 26, 2016
cpp Fix bug in get_max_poly Apr 28, 2016
data Update data checksum Oct 31, 2016
python Update Jun 7, 2017
AUTHORS Remove extra cpp files Apr 27, 2016
CMakeLists.txt Quickfix Apr 28, 2016
LICENSE Remove extra cpp files Apr 27, 2016 Update May 15, 2018 Remove extra cpp files Apr 27, 2016

A newer version of the code is available at DAVIS 2017

A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation (DAVIS)

Package containing helper functions for loading and evaluating DAVIS.

A Matlab version of the same package is also available.


DAVIS (Densely Annotated VIdeo Segmentation), consists of fifty high quality, Full HD video sequences, spanning multiple occurrences of common video object segmentation challenges such as occlusions, motion-blur and appearance changes. Each video is accompanied by densely annotated, pixel-accurate and per-frame ground truth segmentation.


Please cite DAVIS in your publications if it helps your research:

  author    = {Federico Perazzi and
               Jordi Pont-Tuset and
               Brian McWilliams and
               Luc Van Gool and
               Markus Gross and
               Alexander Sorkine-Hornung},
  title     = {A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation},
  booktitle = {The IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
  year      = {2016}

Terms of Use

DAVIS is released under the BSD License [see LICENSE for details]



  • Boost.Python


  • Cython==0.24
  • PyYAML==3.11
  • argparse==1.2.1
  • easydict==1.6
  • future==0.15.2
  • h5py==2.6.0
  • matplotlib==1.5.1
  • numpy==1.11.0
  • prettytable==0.7.2
  • scikit-image==0.12.3
  • scipy==0.17.0



  1. ./ && make -C build/release


  1. pip install virtualenv virtualenvwrapper
  2. source /usr/local/bin/
  3. mkvirtualenv davis
  4. pip install -r python/requirements.txt
  5. export PYTHONPATH=$(pwd)/python/lib
  6. See ROOT/python/lib/davis/ for a list of available options


See source code for documentation.

The directory is structured as follows:

  • ROOT/cpp: Implementation and python wrapper of the temporal stability measure.

  • ROOT/python/tools: contains scripts for evaluating segmentation.

    • : evaluate a technique and store results in HDF5 file
    • read and display evaluation from HDF5.
  • ROOT/python/experiments: contains several demonstrative examples.

  • ROOT/python/lib/davis : library package contains helper functions for parsing and evaluating DAVIS

  • ROOT/data :

    • download input images and annotations.
    • download the CVPR 2016 submission results.