Co-saliency Detection via Co-salient Object Discovery and Recovery

Linwei Ye¹     Zhi Liu¹     Junhao Li¹     Wan-Lei Zhao²     Liquan Shen¹

Image and Video Processing LAB,Shanghai University¹
School of Information Science and Technology,Xiamen University²

 

Abstract

This paper proposes a novel co-saliency model to effectively discover and highlight co-salient objects in a set of images. Based on the gross similarity which combines color features and SIFT descriptors, some co-salient object regions are first discovered in each image as exemplars, which are exploited to generate the exemplar saliency maps with the use of single-image saliency model. Then both local recovery and global recovery of co-salient object regions are performed by propagating the exemplar saliency to the matched regions, and border connectivity is further exploited to generate the region-level co-saliency maps. Finally, the foci of attention area based pixel-level saliency derivation is used to generate the pixel-level co-saliency maps with even better quality. Experimental results on two benchmark datasets demonstrate that the proposed co-saliency model outperforms the state-of-the-art co-saliency models.

 

Proposed Co-Saliency Model

Illustration of the proposed co-saliency model. (a) Input images; (b) co-salient exemplar discovery; (c) exemplar saliency maps; (d) region-level co-saliency maps; (e) pixel-level co-saliency maps.

Results

Comparison with state-of-the-art co-saliency models on iCoseg dataset (top) and MSRC dataset (bottom). From top to bottom: original images, ground truths, co-saliency maps generated using CB [12], HS [17], RFPR [18], SACS [13] and our model, respectively.

 

Quantitative Comparison

>

Precision-recall curves of different models on iCoseg dataset and MSRC dataset.

Citation

L. Ye, Z. Liu, J. Li and L. Shen, "Co-Saliency Detection via Co-Salient Object Discovery and Recovery," Signal Processing Letters, IEEE, vol. 22, no. 11, pp. 2073-2077, Nov 2015.
@ARTICLE{7163310,
author={Linwei Ye and Zhi Liu and Junhao Li and Wan-Lei Zhao and Liquan Shen},
journal={Signal Processing Letters, IEEE},
title={Co-Saliency Detection via Co-Salient Object Discovery and Recovery},
year={2015},
month={Nov},
volume={22},
number={11},
pages={2073-2077},
doi={10.1109/LSP.2015.2458434},
ISSN={1070-9908},} .

 

Downloads

"Co-Saliency Detection via Co-Salient Object Discovery and Recovery" L. Ye, Z. Liu, J. Li and L. Shen, Signal Processing Letters, IEEE, vol. 22, no. 11, pp. 2073-2077, Nov 2015.    [Paper]   [MATLAB Code]