Research map / Adversarial robustness and security / Attacks on text and graphs
Enhancing Tracking Robustness with Auxiliary Adversarial Defense Networks
Zhewei Wu, Ruilong Yu, Qihe Liu, Shuying Cheng, Shilin Qiu, Shijie Zhou
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 |
|---|---|---|---|
| add AADN defense network before the tracker, search branch only | SiamRPN++ · OTB100, CSA attack | Success: 0.349 -> 0.561 | Improved |
| face an attacker adaptive to AADN | AADN · OTB100, CSA attack, search only | Success: 0.561 -> 0.461 | Hurt |
| train with Dua-Loss (cls + reg) rather than classification loss only (ablation) | AADN · OTB100, CSA attack, search only | Success: 0.554 -> 0.561 | Improved |
| train with Dua-Loss rather than regression loss only (ablation) | AADN · OTB100, CSA attack, search only | Success: 0.539 -> 0.561 | Improved |
| apply AADN on the search branch for clean inputs | SiamRPN++ · VOT2018 clean samples | EAO: 0.352 -> 0.245 | Hurt |
| apply AADN on the template branch only for clean inputs | SiamRPN++ · LaSOT clean samples | Success: 0.496 -> 0.501 | No clear effect |
| apply AADN to both branches instead of search only (ablation) | AADN · LaSOT clean samples | Success: 0.432 -> 0.455 | Improved |
| transfer AADN without retraining to SiamRPN, SiamMask, OCEAN | AADN · OTB100, IoU Attack | Success: 0.490 -> 0.517 (SiamRPN) | Improved |
| add AADN on search branch | SiamRPN++ · tracking speed | inference time (ms): 9.35 -> 14.29 | Hurt |
Closest papers by meaning
- Efficient Adversarial Attacks for Visual Object Tracking (ECCV 2020 · similarity 0.90)
- Robust Tracking against Adversarial Attacks (ECCV 2020 · similarity 0.90)
- A Unified Multi-Scenario Attacking Network for Visual Object Tracking (AAAI 2021 · similarity 0.89)
- One-Shot Adversarial Attacks on Visual Tracking With Dual Attention (CVPR 2020 · similarity 0.89)
- IoU Attack: Towards Temporally Coherent Black-Box Adversarial Attack for Visual Object Tracking (CVPR 2021 · similarity 0.88)
- Enhancing Adversarial Transferability with Checkpoints of a Single Model's Training (CVPR 2025 · similarity 0.84)
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.
