Research Analyzer
← Back ICRA 2026

MATrack: Efficient Multiscale Adaptive Tracker for Real-Time Nighttime UAV Operations

Xuzhao Li, Xuchen Li, Shiyu Hu

PDF

AI summary

Key figure (auto-extracted from paper)
MATrack achieves state-of-the-art accuracy and real-time speed for nighttime UAV tracking by dynamically fusing multiscale features and adaptively suppressing background noise.
Nighttime UAV tracking real-time tracking multiscale fusion adaptive token gating low-light vision robotics

Problem

Nighttime UAV tracking struggles with low-light degradation, background clutter, and rapid viewpoint changes, causing existing trackers to drift or fail. Current solutions often introduce visual artifacts, demand excessive computation, or lack long-term stability on resource-constrained platforms.

Approach

MATrack integrates three core modules: a Multiscale Hierarchy Blender to align static and dynamic template features, an Adaptive Key Token Gate to suppress noise and emphasize object cues, and a Nighttime Template Calibrator to ensure stable long-term tracking.

Key results

  • Surpasses SOTA by 5.9% precision, 5.4% normalized precision, and 4.2% AUC on UAVDark135
  • Maintains real-time processing speed of 81 FPS
  • Achieves new state-of-the-art results across five nighttime tracking benchmarks
  • Validated on a real-world UAV platform for reliable nighttime operations

Why it matters

Provides a robust, real-time tracking solution critical for resource-constrained UAV missions like nighttime search and rescue and border patrol.

Abstract

Nighttime UAV tracking faces significant chal- lenges in real-world robotics operations. Low-light conditions not only limit visual perception capabilities, but cluttered backgrounds and frequent viewpoint changes also cause existing trackers to drift or fail during deployment. To address these difficulties, researchers have proposed solutions based on low- light enhancement and domain adaptation. However, these methods still have notable shortcomings in actual UAV systems: low-light enhancement often introduces visual artifacts, domain adaptation methods are computationally expensive and existing lightweight designs struggle to fully leverage dynamic object information. Based on an in-depth analysis of these key issues, we propose MATrack—a multiscale adaptive system designed specifically for nighttime UAV tracking. MATrack tackles the main technical challenges of nighttime tracking through the collaborative work of three core modules: Multiscale Hierarchy Blende (MHB) enhances feature consistency between static and dynamic templates. Adaptive Key Token Gate accurately identifies object information within complex backgrounds. Nighttime Template Calibrator (NTC) ensures stable tracking performance over long sequences. Extensive experiments show that MATrack achieves a significant performance improvement. On the UAVDark135 benchmark, its precision, normalized precision and AUC surpass state-of-the-art (SOTA) methods by 5.9%, 5.4% and 4.2% respectively, while maintaining a real- time processing speed of 81 FPS. Further tests on a real-world UAV platform validate the system’s reliability, demonstrating that MATrack can provide stable and effective nighttime UAV tracking support for critical robotics applications such as nighttime search and rescue and border patrol.

Index terms

Visual Tracking Aerial Systems: Applications Aerial Systems: Perception and Autonomy

Related papers