Research map / Adversarial robustness and security / Attacks on text and graphs
F&F Attack: Adversarial Attack against Multiple Object Trackers by Inducing False Negatives and False Positives
Tao Zhou, Qi Ye, Wenhan Luo, Kaihao Zhang, Zhiguo Shi, Jiming Chen
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| use F&F targeted detection design (erase and inject shifted false alarms) instead of untargeted FP attack (ablation) | ByteTrack attack · MOT17 | IDSWim: 50.17% -> 85.00% | Improved |
| fix k in dynamic-k label assignment (ablation) | F&F attack on ByteTrack · MOT17 | IDSWim: 71.16% -> 85.00% | Improved |
| add regression loss L_L1 to the targeted loss (ablation) | F&F attack on ByteTrack · MOT17 | IDSWim: 64.05% -> 85.00% | Improved |
| attack with F&F instead of Hijacking | CenterTrack (no Kalman filter) · MOT17 | IDSWim: 25.12% -> 74.38% | Improved |
| use higher association threshold in SORT against Daedalus | SORT · MOT17 | IDSWim: Daedalus 51.21% on ByteTrack vs 8.48% on SORT | Hurt |
| increase attacked frames beyond probation period | F&F attack · MOT17 ByteTrack | IDSWim | Improved |
| set shift kappa and scale s so IoU1 slightly above association threshold (ablation) | F&F attack · MOT17 | IDSWim: 85.00% -> 95.02% (ByteTrack, kappa 0.3 s 0.4) | Improved |
| apply spatial smoothing defense with EoT | F&F attack on ByteTrack · MOT17 | IDSWim: 91.4 -> 75.8 | Hurt |
| apply color jitter or Gaussian noise defense | F&F attack on ByteTrack · MOT17 | IDSWim: 91.4 -> 90.8 (CJ), 86.9 (GN) | No clear effect |
Closest papers by meaning
- Fooling Detection Alone is Not Enough: Adversarial Attack against Multiple Object Tracking (ICLR 2020 · similarity 0.89)
- PapMOT: Exploring Adversarial Patch Attack against Multiple Object Tracking (ECCV 2024 · similarity 0.87)
- Out of Sight, Out of Track: Adversarial Attacks on Propagation-based Multi-Object Trackers via Query State Manipulation (CVPR 2026 · similarity 0.87)
- Multiple Object Tracking as ID Prediction (CVPR 2025 · similarity 0.83)
- DeconfuseTrack: Dealing with Confusion for Multi-Object Tracking (CVPR 2024 · similarity 0.82)
- NumbOD: A Spatial-Frequency Fusion Attack Against Object Detectors (AAAI 2025 · similarity 0.81)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
