Light Siamese Network for Long-Term Onboard Aerial Tracking

Abstract

The scarce onboard computational resources and real-time demand restrict the deployment of aerial trackers with sophisticated structures and customized operators. Meanwhile, aerial trackers need updating modules to adapt to continuous appearance variations in real-world long-term scenarios. However, frequent updating will introduce noisy templates and lead to tracking drifts and efficiency drops. Therefore, in this work, we develop a lightweight and highly efficient Siamese tracker for long-term onboard aerial tracking applications. Firstly, we build a compact, plain and deployment-friendly Siamese network based on re-parameterization as the baseline short-term (ST) tracker. Furthermore, we propose a tracking-specific decoupled knowledge distillation guided by strict teacher to unleash the appearance representation potential of the feature extractor without extra inference cost. Specifically, before distillation, the teacher conducts qualification verification to avoid misguiding the student. Then, hard negative background regions are mined and decoupled with the target region, encouraging the student to focus more on similar distractors and informative areas. Finally, to realize efficient and high-confidence long-term (LT) tracking, we design two extensions and incorporate them into the boosted short-term tracker: an initial-template-driven template updater with a corresponding pair-generating strategy to alleviate appearance pollution, and a confidence estimating branch to determine whether to update. Extensive results on large-scale drone benchmarks indicate that our proposed tracker significantly outperforms state-of-the-art aerial trackers. Real-world tests on our customized drone-captured long-term dataset also validate its favorable practicability with a real-time speed of 44 fps on the Lynix KA200 chip.