Air UAV tracking algorithm based on Transformer

In order to solve the problem of target occlusion and long-term tracking in the process of UAV tracking, this paper proposes an aerial UAV tracking algorithm based on Transformer, and uses Transformer encoder to design masks, control four interaction modes, shield four interactions respectively, and obtain a better network model to improve the tracking performance of the algorithm. At the same time, the network framework of this paper introduces a re-search mechanism. When the tracking result is lost or occluded, the target is repositioned by expanding the search range to achieve robust target tracking. The aerial drone tracking algorithm based on Transformer proposed in this paper shows good tracking results on the two data sets of GOT-10k and DTB70. Compared with other algorithms, the algorithm studied in this paper has high accuracy and success rate in tracking UAVs.