HSPT

Learning Discriminative Hidden Structural Parts for Visual Tracking

Longyin Wen, Zhaowei Cai, Dawei Du, Zhen Lei, Stan Z. Li

Abstract

Part-based visual tracking is attractive in recent years due to its robustness to occlusion and non-rigid motion. However, how to automatically generate the discriminative structural parts and consider their relations jointly to construct a more robust tracker still remains unsolved. This paper proposes a discriminative structural part learning method while integrating the structure information, to address the visual tracking problem. Particulary, the state (e.g. position, width and height) of each part is regarded as a hidden variable and inferred automatically by considering the inner structure information of the target and the appearance difference between the target and the background. The inner structure information considering the relationship between neighboring parts, is integrated using a graph model based on a dynamically constructed pair-wise Markov Random Field. Finally, we adopt Metropolis-Hastings algorithm integrated with the online Support Vector Machine to complete the hidden variable inference task. The experimental results on various challenging sequences demonstrate the favorable performance of the proposed tracker over the state-of-the-art ones.

Framework

The proposed method learns the discriminative parts automatically by integrating the structure and discriminative information. In the learning step, both the appearance of the parts and the relationships between them are considered. Since the discriminative parts are not located in the fixed location to the target center, we regard the state of them as hidden variables in the objective function. Then, the objective is optimized by the Metropolis-Hastings (MH) algorithm integrated with the online Support Vector Machine (SVM) method iteratively. In order to achieve more robust performance in complex environments, the bounding box based appearance of the target is also incorporated in our tracker.

Illustrative Results

Downloads

• C++ source code [HSPT].

Citations

If you use the datasets, our tracking results or the source code, please cite our paper:

• Longyin Wen, Zhaowei Cai, Dawei Du, Zhen Lei, and Stan Z Li, " Learning Discriminative Hidden Structural Parts for Visual Tracking", International Workshop on Feature and Similarity Learning for Computer Vision in Conjunction with Asian Conference on Computer Vision (ACCV Workshop), Oral, 2014. [PDF]