Precision in pursuit: a multi-consistency joint approach for infrared anti-UAV tracking

In addition to addressing issues commonly encountered in generic visual object tracking, this task endeavors necessitates coping with the diminutive characteristics of infrared UAV targets, potential discontinuities of targets between adjacent frames due to irregular motion patterns and camera rotations, and the occurrence of drastic appearance deformations. To effectively tackle these challenges, we present a multi-consistency joint approach tailored for infrared anti-UAV tracking. The concept of consistency is realized across three pivotal perspectives: Firstly, by acknowledging the inherent motion consistency of targets, we introduce the Robust Motion Constraint Module (RMCM). This module effectively mitigates tracking drift issues by imposing constraints on the position and size variations of tracked results through an adaptive cost matrix. Secondly, we design the Flexible Spatial Remapping Module (FSRM) grounded in spatial consistency. This module facilitates the identification of correspondences between images in adjacent frames, effectively resolving the substantial target displacement. Lastly, we propose the adaptive template update strategy to address the challenge of significant target deformations. This strategy dynamically selects alternate templates to ensure feature consistency, thereby maintaining stable tracking even in the face of drastic pose changes. The experimental results on benchmark datasets demonstrate that our approach achieves promising advancements compared to state-of-the-art models. Specifically, our approach achieves 69.65 Acc, 80.45 Acc, 65.88 Acc and 50.28 Acc^∗ on the 1st anti-UAV test set dataset, 1st anti-UAV test-dev dataset, 2nd anti-UAV test-dev dataset and 3rd anti-UAV validation set dataset, respectively. We will make our code publicly available upon the acceptance of our paper.