Home


Research


Publications


Downloads


Teaching


Links

Robust Visual Tracking Based on Dynamic Group Sparsity

.Introduction

Based on the observations that a target can be reconstructed from several templates, and only some of the features with discriminative power are significant to separate the target from the background, we propose a novel online tracking algorithm with two stage sparse optimization to jointly
minimize the target reconstruction error and maximize the discriminative power. As the target template and discriminative features usually have temporal and spatial relationship, dynamic group sparsity (DGS) is utilized in our algorithm. The proposed method is compared with three state-of-art trackers using five public challenging sequences, which exhibit appearance changes, heavy occlusions, and pose variations. Our algorithm is shown to outperform these methods.


Reference

Baiyang Liu, Junzhou Huang, Casimir Kulikowski, Lin Yang, "Robust Visual Tracking Using Local Sparse Appearance Model and K-Selection", IEEE Transactions on Pattern Analysis and Machine Intelligence, Volume 35, Issue 12, pp. 2968-2981, December 2013.

Baiyang Liu, Lin Yang, Junzhou Huang, Peter Meer, Leiguang Gong, Casimir Kulikowski, ”Robust and Fast Collaborative Tracking with Two Stage Sparse Optimization”, The 11th European Conference on Computer Vision, Crete, Greece, September, 2010. [PDF]

Junzhou Huang, Xiaolei Huang, Dimitris Metaxas, ”Learning with Dynamic Group Sparsity”, The 12th International Conference on Computer Vision, Kyoto, Japan, October 2009. [SLIDES] [CODE]

Notice: The codes was tested on Windows and MATLAB 2008. If you have any suggestions or you have found a bug, please contact us via email at jzhuang@uta.edu



Figure 1 The tracking results of a car sequence in an open road environment. The vehicle was driven beneath a bridge which led to large illumination changes. Results from our algorithm, MIL, L1, and IVT are given in the first, second, third, and fourth row, respectively.

 

Figure 2 The tracking results of a moving face sequence, which has large pose variation, scaling, and illumination changes. Results from our algorithm, MIL, L1, and IVT are given in the first, second, third, and fourth row, respectively.

 

Figure 3 The tracking results of a plush toy moving around under different pose and illumination conditions. Results from our algorithm, MIL, L1, and IVT are given in the first, second, third, and fourth row, respectively.

 

Figure 4 The tracking results of a face sequence, which includes a lot of pose variations, partial or full occlusions. Results from our algorithm, MIL, L1, and IVT are given in the first, second, third, and fourth row, respectively.

Related Sources

Incremental Learning for Visual Tracking

Tracking with Multiple Instance Learning

L1 Tracker