DeTracker: Motion-decoupled Vehicle Detection and Tracking in Unstabilized Satellite Videos

2026-01-14

AI Summary

This work addresses the performance degradation in multi-object tracking caused by platform jitter and small target sizes in non-stationary satellite videos. To tackle this challenge, the authors propose DeTracker, a novel framework that disentangles global platform motion from local object motion through a global–local motion decoupling mechanism. Furthermore, a temporally dependent feature pyramid is introduced to enable cross-frame feature fusion, significantly enhancing the representation capability and trajectory stability of small targets. The study also constructs SDM-Car-SU, a new benchmark dataset incorporating multi-directional and multi-velocity perturbations. Experimental results demonstrate that DeTracker achieves MOTA scores of 61.1% on the proposed dataset and 47.3% on real-world satellite video sequences, substantially outperforming existing methods.

Technology Category

• Computer Vision: Motion & Tracking
• Machine Learning: Multi-instance/Multi-view Learning
• Intelligent Robots: Motion and Path Planning

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Web Mining and Content Analysis: Topic discovery and tracking
• Security and Privacy: Tracking, profiling, and countermeasures against them

Abstract

Satellite videos provide continuous observations of surface dynamics but pose significant challenges for multi-object tracking (MOT), especially under unstabilized conditions where platform jitter and the weak appearance of tiny objects jointly degrade tracking performance. To address this problem, we propose DeTracker, a joint detection-and-tracking framework tailored for unstabilized satellite videos. DeTracker introduces a Global--Local Motion Decoupling (GLMD) module that explicitly separates satellite platform motion from true object motion through global alignment and local refinement, leading to improved trajectory stability and motion estimation accuracy. In addition, a Temporal Dependency Feature Pyramid (TDFP) module is developed to perform cross-frame temporal feature fusion, enhancing the continuity and discriminability of tiny-object representations. We further construct a new benchmark dataset, SDM-Car-SU, which simulates multi-directional and multi-speed platform motions to enable systematic evaluation of tracking robustness under varying motion perturbations. Extensive experiments on both simulated and real unstabilized satellite videos demonstrate that DeTracker significantly outperforms existing methods, achieving 61.1% MOTA on SDM-Car-SU and 47.3% MOTA on real satellite video data.

Problem

• unstabilized satellite videos
• multi-object tracking
• platform jitter
• tiny objects
• motion perturbations

Innovation

• motion decoupling
• satellite video tracking
• temporal feature fusion
• tiny object tracking
• unstabilized video