Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

DARTer: Dynamic Adaptive Representation Tracker for Nighttime UAV Tracking

Xuzhao Li, Xuchen Li, Shiyu Hu

TL;DR

DARTer tackles nighttime UAV tracking by addressing illumination and viewpoint variability with two core modules: a Dynamic Feature Blender that fuses multi-view features from static and dynamic templates, and a Dynamic Feature Activator that adaptively selects Vision Transformer blocks to reduce redundant computation. The approach pairs these components with a simple prediction head and joint loss, enabling end-to-end training without heavy multi-task objectives. Across five nighttime UAV benchmarks, DARTer achieves state-of-the-art accuracy and real-time performance, demonstrating robust tracking under extreme conditions and practical applicability for real-world operations. This work advances adaptive feature modeling and lightweight transformer-based tracking in harsh lighting, offering a promising direction for low-light visual perception in UAV systems.

Abstract

Nighttime UAV tracking presents significant challenges due to extreme illumination variations and viewpoint changes, which severely degrade tracking performance. Existing approaches either rely on light enhancers with high computational costs or introduce redundant domain adaptation mechanisms, failing to fully utilize the dynamic features in varying perspectives. To address these issues, we propose \textbf{DARTer} (\textbf{D}ynamic \textbf{A}daptive \textbf{R}epresentation \textbf{T}racker), an end-to-end tracking framework designed for nighttime UAV scenarios. DARTer leverages a Dynamic Feature Blender (DFB) to effectively fuse multi-perspective nighttime features from static and dynamic templates, enhancing representation robustness. Meanwhile, a Dynamic Feature Activator (DFA) adaptively activates Vision Transformer layers based on extracted features, significantly improving efficiency by reducing redundant computations. Our model eliminates the need for complex multi-task loss functions, enabling a streamlined training process. Extensive experiments on multiple nighttime UAV tracking benchmarks demonstrate the superiority of DARTer over state-of-the-art trackers. These results confirm that DARTer effectively balances tracking accuracy and efficiency, making it a promising solution for real-world nighttime UAV tracking applications.

DARTer: Dynamic Adaptive Representation Tracker for Nighttime UAV Tracking

TL;DR

DARTer tackles nighttime UAV tracking by addressing illumination and viewpoint variability with two core modules: a Dynamic Feature Blender that fuses multi-view features from static and dynamic templates, and a Dynamic Feature Activator that adaptively selects Vision Transformer blocks to reduce redundant computation. The approach pairs these components with a simple prediction head and joint loss, enabling end-to-end training without heavy multi-task objectives. Across five nighttime UAV benchmarks, DARTer achieves state-of-the-art accuracy and real-time performance, demonstrating robust tracking under extreme conditions and practical applicability for real-world operations. This work advances adaptive feature modeling and lightweight transformer-based tracking in harsh lighting, offering a promising direction for low-light visual perception in UAV systems.

Abstract

Nighttime UAV tracking presents significant challenges due to extreme illumination variations and viewpoint changes, which severely degrade tracking performance. Existing approaches either rely on light enhancers with high computational costs or introduce redundant domain adaptation mechanisms, failing to fully utilize the dynamic features in varying perspectives. To address these issues, we propose \textbf{DARTer} (\textbf{D}ynamic \textbf{A}daptive \textbf{R}epresentation \textbf{T}racker), an end-to-end tracking framework designed for nighttime UAV scenarios. DARTer leverages a Dynamic Feature Blender (DFB) to effectively fuse multi-perspective nighttime features from static and dynamic templates, enhancing representation robustness. Meanwhile, a Dynamic Feature Activator (DFA) adaptively activates Vision Transformer layers based on extracted features, significantly improving efficiency by reducing redundant computations. Our model eliminates the need for complex multi-task loss functions, enabling a streamlined training process. Extensive experiments on multiple nighttime UAV tracking benchmarks demonstrate the superiority of DARTer over state-of-the-art trackers. These results confirm that DARTer effectively balances tracking accuracy and efficiency, making it a promising solution for real-world nighttime UAV tracking applications.
Paper Structure (10 sections, 4 equations, 2 figures, 4 tables)

This paper contains 10 sections, 4 equations, 2 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Methods
  3. Dark Feature Blender
  4. Dynamic Feature Activator
  5. Prediction Head and Training Loss
  6. Experiment
  7. Implementation Details
  8. Comparison Results
  9. Ablation Study and Analysis
  10. Conclusion

Figures (2)

  • Figure 1: (a) Overview architecture of DARTer. The nighttime dynamic features of the static and dynamic templates are fused. The ViT blocks are dynamically activated according to the currently extracted nighttime features. (b) Diagram of Dynamic Feature Activator. The DFA module performs token extraction, transforms them through linear and convolution operations, and then conducts an activation process to adaptively select ViT layers and improve efficiency.
  • Figure 2: Qualitative comparison results of our tracker with other two latest trackers (i.e., DCPT zhu2024dcpt and AVTrack lilearningicml in representative nighttime scenarios. Better viewed in color with zoom-in.