Structural target-aware model for thermal infrared tracking

Thermal InfraRed (TIR) target trackers are easy to be interfered by similar objects, while susceptible the influence of occlusion. To solve these problems, we propose a structural target-aware model (STAMT) for thermal infrared tracking tasks. Specifically, proposed STAMT tracker can learn model, which add more attention area accurately identify from objects. In addition, considering situation that is partially occluded in process, weight locate through unoccluded reliable part. Ablation studies show effectiveness each component tracker. Without bells and whistles, experimental results demonstrate our performs favorably against state-of-the-art on PTB-TIR LSOTB-TIR datasets.