Joint Segmentation and Pose Tracking of Human in Natural Videos

Taegyu Lim, Seunghoon Hong, Bohyung Han, Joon Hee Han; Proceedings of the IEEE International Conference on Computer Vision (ICCV), 2013, pp. 833-840

Abstract


We propose an on-line algorithm to extract a human by foreground/background segmentation and estimate pose of the human from the videos captured by moving cameras. We claim that a virtuous cycle can be created by appropriate interactions between the two modules to solve individual problems. This joint estimation problem is divided into two subproblems, foreground/background segmentation and pose tracking, which alternate iteratively for optimization; segmentation step generates foreground mask for human pose tracking, and human pose tracking step provides foreground response map for segmentation. The final solution is obtained when the iterative procedure converges. We evaluate our algorithm quantitatively and qualitatively in real videos involving various challenges, and present its outstanding performance compared to the state-of-the-art techniques for segmentation and pose estimation.

Related Material


[pdf]
[bibtex]
@InProceedings{Lim_2013_ICCV,
author = {Lim, Taegyu and Hong, Seunghoon and Han, Bohyung and Han, Joon Hee},
title = {Joint Segmentation and Pose Tracking of Human in Natural Videos},
booktitle = {Proceedings of the IEEE International Conference on Computer Vision (ICCV)},
month = {December},
year = {2013}
}