Research map / Adversarial robustness and security / Attacks on text and graphs
Fractal Camouflage: A Bio-Inspired Approach for Multi-Scale Adversarial Attacks in the Infrared Domain
Chengyin Hu, Xin Wang, Rui Qiu, Zhe Jia, Yingying Zhao, Kai Wang, Xu Kang, Yiwei Wei
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 |
|---|---|---|---|
| increase fractal depth from 1 to 3 in the physical attack (ablation) | AdvFractal · Physical attack on YOLOv3 | average ASR (%): 82.87 -> 97.54 | Improved |
| increase fractal depth from 1 to 3 at the farthest distance (ablation) | AdvFractal · Physical attack on YOLOv3, 9.4m | ASR (%): 0.01 -> 98.67 | Improved |
| use progressive halving of layer width with PSO instead of equal widths (ablation) | AdvFractal · Physical attack on YOLOv3 | average ASR (%): 19.72 -> 97.54 | Improved |
| use PSO instead of random optimization with halved widths (ablation) | AdvFractal · Physical attack on YOLOv3 | average ASR (%): 66.34 -> 97.54 | Improved |
| use PSO instead of GA or DE for optimization (ablation) | AdvFractal · FLIR digital, YOLOv3 | ASR / query count | No clear effect |
| increase fractal depth in digital attacks (ablation) | AdvFractal (PSO) · FLIR digital, YOLOv3, length 0.2 | ASR (%): 70.06 -> 85.99 | Improved |
| increase initial layer length ratio (ablation) | AdvFractal (PSO, depth 1) · FLIR digital, YOLOv3 | ASR (%): 70.06 -> 99.36 | Improved |
| increase PSO iterations beyond 30 (ablation) | AdvFractal · FLIR digital, YOLOv3 | ASR (%): 98.09 -> 99.36 | No clear effect |
| target detector trained with adversarial training | AdvFractal · FLIR digital, YOLOv3 | ASR (%): 99.36 -> 59.24 | Hurt |
| target detector uses digital watermarking defense | AdvFractal · FLIR digital, YOLOv3 | ASR (%): 99.36 -> 64.52 | Hurt |
| attack different detector architectures | AdvFractal · FLIR digital, ten detectors | ASR (%): 100 on YOLOF -> 59.60 on RetinaNet | Mixed |
Closest papers by meaning
- Unleashing the Representational Power of Fourier Shapes for Attacking Infrared Object Detection (ICML 2026 · similarity 0.86)
- Unified Adversarial Patch for Visible-Infrared Cross-Modal Attacks in the Physical World (TPAMI 2023 · similarity 0.86)
- Full-Distance Evasion of Pedestrian Detectors in the Physical World (NeurIPS 2024 · similarity 0.84)
- Adversarial Attacks on Event-Based Pedestrian Detectors: A Physical Approach (AAAI 2025 · similarity 0.83)
- Infrared Adversarial Car Stickers (CVPR 2024 · similarity 0.82)
- PhyCamo: A Robust Physical Camouflage via Contrastive Learning for Multi-View Physical Adversarial Attack (AAAI 2025 · similarity 0.82)
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.
