Template Enhancement and Mask Generation for Siamese Tracking

Siamese tracking methods have become the focus of visual tracking in recent years. Advanced Siamese trackers perform well on certain benchmarks, but there are still some limitations. First, most Siamese trackers adopt the initial frame as a single template, which leads to underfitting and reduces the ability to predict instances. Second, mainstream trackers report a rectangular bounding box as a prediction, resulting in poor accuracy of non-rigid objects. Therefore, we propose the template enhancement and mask generation for Siamese tracking. Given that the essence of Siamese trackers is instance learning, we propose constructing an alternative template explicitly to address the underfitting of the instance space. Moreover, in order to improve the tracking accuracy, we obtain the descriptor aggregation to transform the semantic segmentation outputs for mask prediction. Finally, we propose the SiamEM through the fusion of the above approaches. Comprehensive experiments show that template enhancement and mask generation significantly improve Siamese trackers on benchmarks.

I. Introduction

Visual object tracking is a basic and important task in the field of computer vision. For a video or an image sequence, given a region of interest (usually labeled by a rectangular bounding box (bbox)), the target position should be predicted in subsequent frames. The visual tracking task is usually a component of a large-scale visual system, so the target of which can be arbitrary; that is, the category of the target is a priori unknown to the tracking algorithm.