Hierarchical Saliency Detection

Qiong Yan      Li Xu      Jianping Shi     Jiaya Jia

The Chinese Univeristy of Hong Kong


An overview of our hierarchical framework. We extract three image layers from the input, and then compute saliency cues from each of these layers. They are finally fed into a hierarchical model to get the final results.


When dealing with objects with complex structures, saliency detection confronts a critical problem – namely that detection accuracy could be adversely affected if salient foreground or background in an image contains small-scale high-contrast patterns. This issue is common in natural images and forms a fundamental challenge for prior methods. We tackle it from a scale point of view and propose a multi-layer approach to analyze saliency cues. The final saliency map is produced in a hierarchical model. Different from varying patch sizes or downsizing images, our scale-based region handling is by finding saliency values optimally in a tree model. Our approach improves saliency detection on many images that cannot be handled well traditionally. A new dataset is also constructed.



All results of our method on MSRA-1000, CSSD and ECSSD.


Quantitative Comparison


Precision-recall comparison with state-of-the-art method. For details on the abbreviations of methods please refer to our paper. Our method is the top orange curve.



Snapshot for paper "Hierarchical Saliency Detection"
Qiong Yan, Li Xu, Jianping Shi, Jiaya Jia
IEEE Conference on Computer Vision and Pattern Recognition (CVPR), 2013
   [Paper (pdf, 2.06MB)]
   [Supplementary file (pdf, 2.40MB)]
    [ECSSD] (NEW)

Other links



Last update: April 21, 2013