Making the Right Moves: Guiding Alpha-Expansion Using Local Primal-Dual Gaps

Abstract

This paper presents a new adaptive graph-cut based move-making algorithm for energy minimization. Traditional move-making algorithms such as Expansion and Swap operate by searching for better solutions in some predefined moves spaces around the current solution. In contrast, our algorithm uses the primal-dual interpretation of the Expansion-move algorithm to adaptively compute the best move-space to search over. At each step, it tries to greedily find the move-space that will lead to biggest decrease in the primal-dual gap. We test different variants of our algorithm on a variety of image labelling problems such as object segmentation and stereo. Experimental results show that our adaptive strategy significantly outperforms the conventional Expansion-move algorithm, in some cases cutting the runtime by 50%.

Cite

Text

Batra and Kohli. "Making the Right Moves: Guiding Alpha-Expansion Using Local Primal-Dual Gaps." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2011. doi:10.1109/CVPR.2011.5995449

Markdown

[Batra and Kohli. "Making the Right Moves: Guiding Alpha-Expansion Using Local Primal-Dual Gaps." IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2011.](https://mlanthology.org/cvpr/2011/batra2011cvpr-making/) doi:10.1109/CVPR.2011.5995449

BibTeX

@inproceedings{batra2011cvpr-making,
  title     = {{Making the Right Moves: Guiding Alpha-Expansion Using Local Primal-Dual Gaps}},
  author    = {Batra, Dhruv and Kohli, Pushmeet},
  booktitle = {IEEE/CVF Conference on Computer Vision and Pattern Recognition},
  year      = {2011},
  pages     = {1865-1872},
  doi       = {10.1109/CVPR.2011.5995449},
  url       = {https://mlanthology.org/cvpr/2011/batra2011cvpr-making/}
}