| CVPR2013 |
|
Maximum Cohesive Grid of Superpixels for Fast Object Localization |
|
Liang Li Wei Feng Liang Wan Jiawan Zhang Tien-Tsin Wong |
|
Algorithm flow of maximum cohesive grid regularization of arbitrary superpixels. |
|
Abstract This paper addresses a challenging problem of regular- izing arbitrary superpixels into an optimal grid structure, which may significantly extend current low-level vision al- gorithms by allowing them to use superpixels (SPs) conve- niently as using pixels. For this purpose, we aim at con- structing maximum cohesive SP-grid, which is composed of real nodes, i.e. SPs, and dummy nodes that are meaningless in the image with only position-taking function in the grid. For a given formation of image SPs and proper number of dummy nodes, we first dynamically align them into a grid based on the centroid localities of SPs. We then define the SP-grid coherence as the sum of edge weights, with SP lo- cality and appearance encoded, along all direct paths con- necting any two nearest neighboring real nodes in the grid. We finally maximize the SP-grid coherence via cascade dy- namic programming. Our approach can take the regional objectness as an optional constraint to produce more se- mantically reliable SP-grids. Experiments on object local- ization show that our approach outperforms state-of-the-art methods in terms of both detection accuracy and speed. We also find that with the same searching strategy and features, object localization at SP-level is about 100-500 times faster than pixel-level, with usually better detection accuracy. |
||||
|
|
Paper (PDF, 1.9M) |
|||
|
BibTex:
@article{Gridization2013, |
||||
