Research map / Adversarial robustness and security / Attacks on text and graphs
A Unified Multi-Scenario Attacking Network for Visual Object Tracking
Xuesong Chen, Canmiao Fu, Feng Zheng, Yong Zhao, Hongsheng Li, Ping Luo, Guo-Jun Qi
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 UEN non-targeted perturbations instead of iterative attack methods | SiamRPN and SiamRPN++ trackers · OTB100 | Suc/Pre (%): BIM 4.6/4.6 -> UEN 3.6/2.9; BIM 11.9/13.7 -> UEN 0.3/0.3; BIM 33.7/44.8 -> UEN 8.9/10.5 | Improved |
| Use UEN non-targeted perturbations | DiMP (ResNet18) · OTB100, UAV123, and GOT10K | Capture/Confidence: OTB100: 0.0/0.1; UAV123: 0.0/0.0; GOT10K: 0.0/0.0 | Improved |
| Generate perturbations with UEN rather than iterative baselines | SiamRPN-based trackers · OTB100 | Time (ms): 2.25 ms; BIM 99.3 ms, 225.5 ms, and 1249.4 ms across the three listed trackers | Improved |
| Transfer UEN perturbations between tracker models | SiamRPN-based trackers · OTB100 | Success-rate drop/Precision-rate drop: 59.9/58.5; 62.0/60.0 | Improved |
| Use UEN targeted perturbations to hijack trajectory or alter box shape | SiamRPN-based trackers · OTB100, UAV123, and GOT10K | TR(10)/TR(15)/SR (%): OTB100: 35.6/86.9/55.9, 74.2/83.1/45.3, 37.6/55.7/58.8 | Improved |
| Apply a general adversarial patch to the target | SiamRPN (Alex) · OTB100 | Suc/Pre (%): 66.6/87.6 -> 19.4/25.6 | Improved |
| Apply a general adversarial patch to the target | DiMP (ResNet18) · OTB100 | Suc/Pre (%): 67.6/87.2 -> 44.1/78.3 | Improved |
Closest papers by meaning
- Efficient Adversarial Attacks for Visual Object Tracking (ECCV 2020 · similarity 0.90)
- IoU Attack: Towards Temporally Coherent Black-Box Adversarial Attack for Visual Object Tracking (CVPR 2021 · similarity 0.89)
- Enhancing Tracking Robustness with Auxiliary Adversarial Defense Networks (ECCV 2024 · similarity 0.89)
- One-Shot Adversarial Attacks on Visual Tracking With Dual Attention (CVPR 2020 · similarity 0.89)
- SPARK: Spatial-aware Online Incremental Attack Against Visual Tracking (ECCV 2020 · similarity 0.88)
- Robust Tracking against Adversarial Attacks (ECCV 2020 · similarity 0.88)
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.
