Salient Object Detection for Searched Web Images via Global Saliency

Peng Wang1, Jingdong Wang2, Gang Zeng1, Jie Feng1, Hongbin Zha1 and Shipeng Li2
1Key Laboratory on Machine Perception, Peking University  2 Microsoft Research Asia
 

Abstract

In this paper, we deal with the problem of detecting the existence and the location of salient objects for thumbnail images on which most search engines usually perform visual analysis in order to handle web-scale images. Different from previous techniques, such as sliding window-based or segmentation-based schemes for detecting salient objects, we propose to use a learning approach, random forest in our solution. Our algorithm exploits global features from multiple saliency indicators to directly predict the existence and the position of the salient object. To validate our algorithm, we constructed a large image database collected from Bing image search, that contains hundreds of thousands of manually labeled web images.


Framework

Overview
 
 

Download Paper

| Paper (PDF) |

Download Salient Object Dataset

| The searched web image dataset (Webpage)|
 

Bibtex

@inproceedings{WangWZFZLCVPR12,
 author = "Peng Wang and Jingdong Wang and Gang Zeng and Jie Feng and Hongbin Zha and Shipeng Li",
 title  = "Salient Object Detection for Searched Web Images via Global Saliency",
 booktitle = "CVPR",
 year   = "2012"
 }
 

Results

The object localization Results
 
Examples
Example1
Object localization results on the MSRA B salient object database
Example1
Object localization results on the web image database
 
 

Reference

Huaizu Jiang, Jingdong Wang, Zejian Yuan, Tie Liu, Nanning Zheng, Shipeng Li. Automatic Salient Object Segmentation Based on Context and Shape Prior. British Machine Vision Conference (BMVC) 2011.
 
HTML Hit Counters