M. Grundmann, V. Kwatra, M. Han, I. A. Essa. Efficient hierarchical graph-based video segmentation. Conference on Computer Vision and Pattern, San Francisco, 2010.

Grundmann et al. extend the graph-based image segmentation algorithm proposed by Felzenswalb and Huttenlocher [1] to video and also provide an efficient implementation able to process arbitrarily long videos on GitHub. They also provide some video sequences without ground truth for qualitative evaluation, see here. The algorithm can also be used for annotation online at videosegmentation.com - the project is currently maintained by Daniel Castro.

Figure 1 shows a segmentation example on the Sintel dataset [2] generated using a custom implementation inspired by the original implementation by Felzenswalb and Huttenlocher [1].

The implementation is available on GitHub.

Figure 1: Example of video segmentation for the alley_1 sequence of the Sintel dataset [2]. Top: original sequence; middle: oversegmentation; bottom: hierarchical segmentation.

Another implementation is also available as part of the libsvx library [3].

  • [1] P. F. Felzenswalb and D. P. Huttenlocher. Efficient graph-based image segmentation. International Journal of Computer Vision, volume 59, number 2, 2004.
  • [2] D. J. Butler, J. Wulff, G. B. Stanley, M. J. Black. A naturalistic open source movie for optical flow evaluation. European Conference on Computer Vision, 2012.
  • [3] C. Xu, J. J. Corso. Evaluation of super-voxel methods for early video processing. Conference on Computer Vision and Pattern Recognition, 2012.
What is your opinion on this article? Let me know your thoughts on Twitter @davidstutz92 or LinkedIn in/davidstutz92.