Problem
Research questions and friction points this paper is trying to address.
Adversarial Attacks
Multi-Object Tracking
Query Propagation
Temporal Memory
Tracking Vulnerability
Out of Sight, Out of Track: Adversarial Attacks on Propagation-based Multi-Object Trackers via Query State Manipulation
🤖 AI Summary
This work addresses the vulnerability of query propagation–based multi-object trackers (TBP) to adversarial attacks, stemming from their reliance on temporal query dependencies. The paper proposes FADE, a novel attack framework that for the first time exposes and exploits the fragility of the query propagation mechanism inherent in TBP architectures. FADE integrates temporal query flooding and sequential memory corruption strategies with a differentiable, physically realizable sensor spoofing simulation to effectively disrupt the tracker’s query budget allocation and state consistency. Experimental results on MOT17 and MOT20 benchmarks demonstrate that FADE significantly degrades the performance of state-of-the-art TBP trackers, inducing substantial identity switches and trajectory fragmentation, thereby confirming its effectiveness and practical threat potential.
📝 Abstract
Recent Tracking-by-Query-Propagation (TBP) methods have advanced Multi-Object Tracking (MOT) by enabling end-to-end (E2E) pipelines with long-range temporal modeling. However, this reliance on query propagation introduces unexplored architectural vulnerabilities to adversarial attacks. We present FADE, a novel attack framework designed to exploit these specific vulnerabilities. FADE employs two attack strategies targeting core TBP mechanisms: (i) Temporal Query Flooding: Generates spurious temporally consistent track queries to exhaust the tracker's limited query budget, forcing it to terminate valid tracks. (ii) Temporal Memory Corruption: Directly attacks the query updater's memory by severing temporal links via state de-correlation and erasing the learned feature identity of matched tracks. Furthermore, we introduce a differentiable pipeline to optimize these attacks for physical-world realizability by leveraging simulations of advanced perception sensor spoofing. Experiments on MOT17 and MOT20 benchmarks demonstrate that FADE is highly effective against state-of-the-art TBP trackers, causing significant identity switches and track terminations.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Adversarial Attacks
Query Propagation
Multi-Object Tracking
Temporal Memory Corruption
Differentiable Optimization
🔎 Similar Papers
No similar papers found.
