Incremental Slow Feature Analysis

Slow Feature Analysis (SFA) is a subspace learning method inspired by the human visual system; however, it is seldom seen in computer vision. Motivated by its application for unsupervised activity analysis, we develop SFA's first implementation of online temporal video segmentation to detect episodes of motion changes.

If this does not load after a short while:
  • Refresh the page
  • Ensure JavaScript is enabled
  • Install the latest Flash Player

We employ a domain-specific indefinite kernel which takes the data representation into account to introduce robustness. As our kernel is indefinite (i.e. defines instead of a Hilbert, a Krein space), we formulate SFA in Krein space. Finally, we propose an incremental kernel SFA framework which utilizes the special properties of our kernel.

The results of our online temporal video segmentation can be seen in the videos above. A graphic of the incremental slow feature analysis is given below. Please refer to the related publication for more detail:

S. Liwicki, S. Zafeiriou, M. Pantic. “Online Kernel Slow Feature Analysis for Temporal Video Segmentation and Tracking”, IEEE Transaction on Image Processing, vol. 24, no. 10, pp. 2955 – 2970, Oct 2015. [pdf | page]

S. Liwicki, S. Zafeiriou, M. Pantic. “Incremental Slow Feature Analysis with Indefinite Kernel for Online Temporal Video Segmentation”, Proceedings of the 11th Asian Conference on Computer Vision (ACCV’12), Daejeon, South Korea, pp. 162 – 176, 2012. [pdf | page | poster]

SFA-workflow

Resources

You can download the Matlab implementation of our Incremental SFA here:

The video sequences were provided by:

Here are the facial expression videos with groundtruth:

000webhost logo