Abstract | ||
---|---|---|
Image saliency and co-saliency detection that aim to detect salient objects in an image or common salient objects in a group of images are import in computer vision. Researchers often treat saliency and co-saliency as two separate problems. In this paper, we show that these two problems can be solved in a single framework, i.e., treating saliency and co-saliency as finding suitable exemplars. Image-level and region-level exemplars are proposed to obtain the similar images and to propagate the saliency values, respectively. Our method only requires a small number of labeled images having similar appearances with a query image. The exemplars help to detect the real salient objects, which is different from the conventional heuristic methods that are fragile for the images with complex scenes. We have conducted abundant experiments on saliency and co-saliency benchmark datasets, which verifies the effectiveness of our method. |
Year | DOI | Venue |
---|---|---|
2020 | 10.1016/j.neucom.2019.09.011 | Neurocomputing |
Keywords | Field | DocType |
Saliency detection,Co-saliency detection,Exemplar,Label propagation | Heuristic,Pattern recognition,Salience (neuroscience),Salient objects,Artificial intelligence,Mathematics | Journal |
Volume | ISSN | Citations |
371 | 0925-2312 | 0 |
PageRank | References | Authors |
0.34 | 0 | 5 |
Name | Order | Citations | PageRank |
---|---|---|---|
Rui Huang | 1 | 61 | 23.21 |
Wei Feng | 2 | 501 | 61.25 |
Zezheng Wang | 3 | 3 | 0.71 |
Yan Xing | 4 | 43 | 15.82 |
Yaobin Zou | 5 | 3 | 2.73 |