Two-stage aware attentional Siamese network for visual tracking

Siamese networks have achieved great success in visual tracking with the advantages of speed and accuracy. However, how to track an object precisely robustly still remains challenging. One reason is that multiple types features are required achieve good precision robustness, which unattainable by a single training phase. Moreover, usually struggle online adaption problem. In this paper, we present novel two-stage aware attentional network for (Ta-ASiam). Concretely, first propose position-aware appearance-aware strategy optimize different layers network. By introducing diverse patterns, two can be captured simultaneously. Then, following rule feature distribution, effective selection module constructed combining both channel spatial attention adapt rapid appearance changes object. Extensive experiments on various latest benchmarks well demonstrated effectiveness our method, significantly outperforms state-of-the-art trackers.