Downloads - DAVIS 2017

Evaluation code, dataset and results


Evaluation Test-Dev and Test-Challenge


The official metrics will be computed using the images and annotations at 480p resolution, but feel free to use the full resolution ones (4k, 1080p, etc.) in any step of your research.
TrainVal - Images and Annotations
Test-Dev - Images and First-Frame Annotations
Test-Challenge - Images and First-Frame Annotations

Object categories

Contains the semantic masks for all the publicly available frames, a JSON file with the category for each object and another JSON file with the id and the super category for each category.
TrainVal, Test-Dev, Test-Challenge


Please cite this paper if you use the code or dataset.

The 2017 DAVIS Challenge on Video Object Segmentation
J. Pont-Tuset, F. Perazzi, S. Caelles, P. Arbeláez, A. Sorkine-Hornung, and L. Van Gool
arXiv:1704.00675, 2017
[PDF] [BibTex]

  author = {Jordi Pont-Tuset and Federico Perazzi and Sergi Caelles and Pablo Arbel\'aez and Alexander Sorkine-Hornung and Luc {Van Gool}},
  title = {The 2017 DAVIS Challenge on Video Object Segmentation},
  journal = {arXiv:1704.00675},
  year = {2017}
Please also consider citing the following paper as some sequences are borrowed from it.

A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation
F. Perazzi, J. Pont-Tuset, B. McWilliams, L. Van Gool, M. Gross, and A. Sorkine-Hornung
Computer Vision and Pattern Recognition (CVPR) 2016
[PDF] [Supplemental] [BibTex]

author = {F. Perazzi and J. Pont-Tuset and B. McWilliams and L. {Van Gool} and M. Gross and A. Sorkine-Hornung},
title = {A Benchmark Dataset and Evaluation Methodology for Video Object Segmentation},
booktitle = {Computer Vision and Pattern Recognition},
year = {2016}