Exploring reliable infrared object tracking with spatio-temporal fusion transformer

Meibin Qi, Qinxin Wang, Shuo Zhuang, Ke Zhang, Kunyuan Li, Yimin Liu, Yanfang Yang

Abstract: Highlights•This paper proposes a spatio-temporal fusion Transformer model for infrared target tracking.•The dynamic template state estimation module integrates the fine-grained feature information of salient points.•The designed IOU-Aware TSE head leverages an IoU-based metric to evaluate the current state of the target image.