Real-Time Salient Object Detection with a Minimum Spanning Tree

Spotlight Presentation in CVPR 2016

Wei-Chih Tu1    Shengfeng He2    Qingxiong Yang2    Shao-Yi Chien1

1Graduate Institute of Electronics Engineering, National Taiwan University
2Department of Computer Science, City University of Hong Kong

An overview of our salient object detection system. 


We present a real-time salient object detection system based on the minimum spanning tree. Due to the fact that background regions are typically connected to the image boundaries, salient objects can be extracted by computing the distances to image boundaries. However, measuring the image boundary connectivity efficiently is a challenging problem. Existing methods either rely on superpixel representation to reduce the processing units or approximate the distance transform. Instead, we propose to measure distance on a minimum spanning tree. The minimum spanning tree representation of an image largely reduces the search space of shortest paths, resulting an efficient and high quality distance transform algorithm. Extensive evaluations show that the proposed algorithm achieves the leading performance compared to the state-of-the-art methods in terms of efficiency and accuracy.


Results of our method on MSRA-1000, ECSSD and PASCAL-S.


    author = {Tu, Wei-Chih and He, Shengfeng and Yang, Qingxiong and Chien, Shao-Yi},
    title = {Real-Time Salient Object Detection with a Minimum Spanning Tree},
    booktitle = {IEEE Conference on Computer Vision and Pattern Recognition (CVPR)},
    month = {June},
    year = {2016},


Evaluation with weighted F-measure. The proposed method outperforms other compared methods on three popularly used datasets.

Sample results from three datasets. Please refer to our paper for more details about the compared methods.

External Links