Research map / Adversarial robustness and security / Attacks on text and graphs
Physical Adversarial Examples for Person Detectors in Thermal Images Based on 3D Modeling
Xiaopei Zhu, Siyuan Huang, Zhanhao Hu, Jianmin Li, Jun Zhu, Xiaolin Hu
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 |
|---|---|---|---|
| optimize black-patch layout instead of random placement | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 38.19% -> 86.38% | Improved |
| optimize layout instead of random placement on physical clothing (indoor) | aerogel clothing · Physical, indoor | ASR: 26.53% -> 80.11% | Improved |
| optimize layout instead of random placement on physical clothing (outdoor) | aerogel clothing · Physical, outdoor | ASR: 23.03% -> 76.85% | Improved |
| increase number of patches from 6 to 36 (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 22.38% -> 86.38% | Improved |
| increase number of patches from 36 to 60 (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 86.38% -> 83.73% | Hurt |
| increase patch size from 2 cm to 6 cm (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 51.30% -> 86.38% | Improved |
| increase patch size from 6 cm to 12 cm (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 86.38% -> 78.12% | Hurt |
| add distance loss L_dist (lambda 0 -> 0.1) (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 72.66% -> 86.38% | Improved |
| raise lambda from 0.1 to 1 (ablation) | 3D infrared clothing · FLIR_ADAS_1_3 digital | ASR: 86.38% -> 76.61% | Hurt |
| optimize against an ensemble of detectors instead of YOLOv9 only (ablation) | 3D infrared clothing · Transfer to DETR, digital | ASR: 29.75% -> 81.03% | Improved |
| optimize against an ensemble of detectors instead of YOLOv9 only (ablation) | 3D infrared clothing · Transfer to YOLOv8, digital | ASR: 58.36% -> 80.28% | Improved |
| optimize against an ensemble of detectors instead of YOLOv9 only (ablation) | aerogel clothing · Transfer to DETR, physical | ASR: 21.69% -> 70.05% | Improved |
Closest papers by meaning
- Infrared Invisible Clothing: Hiding From Infrared Detectors at Multiple Angles in Real World (CVPR 2022 · similarity 0.94)
- Infrared Adversarial Car Stickers (CVPR 2024 · similarity 0.87)
- Adversarial Attacks on Event-Based Pedestrian Detectors: A Physical Approach (AAAI 2025 · similarity 0.86)
- DAP: A Dynamic Adversarial Patch for Evading Person Detectors (CVPR 2024 · similarity 0.85)
- AdvDisplay: Adversarial Display Assembled by Thermoelectric Cooler for Fooling Thermal Infrared Detectors (AAAI 2025 · similarity 0.84)
- UV-Attack: Physical-World Adversarial Attacks on Person Detection via Dynamic-NeRF-based UV Mapping (ICLR 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.
