Research map / Adversarial robustness and security / Attacks on text and graphs
Deformable and Multi-view Gradient-Aligned Physical Adversarial Camouflage
Jiawei Liang, Puning Zhao, tianrui lou, Haoqing Zhang, Xianghao Jiao, Bozheng Lin, Ming Zhang, Xiaochun Cao
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 Deformable Texture Field (DTF) parameterization (ablation) | adversarial camouflage · YOLOv3 0 deg | AP@0.5 (%): 61.70 -> 33.06 | Improved |
| add Gradient-Aligned Meta-Optimization (GAMO) (ablation) | adversarial camouflage · YOLOv3 0 deg | AP@0.5 (%): 61.70 -> 40.13 | Improved |
| combine DTF and GAMO (ablation) | adversarial camouflage · YOLOv3 0 deg | AP@0.5 (%): 33.06 -> 19.05 | Improved |
| combine DTF and GAMO (ablation) | adversarial camouflage · Faster R-CNN 0 deg (black-box) | AP@0.5 (%): 85.17 -> 42.92 | Improved |
| use GAMO instead of PCGrad for gradient conflicts (ablation) | DTF · Gradient-conflict handling | AP@0.5 (%): 20.25 -> 7.68 | Improved |
| use GAMO instead of GRAC for gradient conflicts (ablation) | DTF · Gradient-conflict handling | AP@0.5 (%): 12.34 -> 7.68 | Improved |
| add GAMO to DTF (ablation) | DTF · Gradient cosine similarity | gradient cosine similarity: 0.0048 -> 0.0070 | Improved |
| proposed method vs GRAC on GLIP | Ours · GLIP transfer | AP@0.5 (%): 41.42 -> 43.72 | Hurt |
| proposed method vs GRAC on DINO | Ours · DINO transfer | AP@0.5 (%): 23.03 -> 13.36 | Improved |
| proposed method vs GRAC on physical Faster R-CNN 20 deg | Ours · Physical 1:24 car | AP@0.5 (%): 54.30 -> 41.15 | Improved |
| proposed method vs GRAC per-view cost | Ours · Efficiency per view | latency (ms): 430.10 -> 135.17 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- RAUCA — RAUCA: A Novel Physical Adversarial Attack on Vehicle Detectors via Robust and Accurate Camouflage Generation (ICML 2024) · better on 3 of 3 shared results
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Gradient-Reweighted Adversarial Camouflage for Physical Object Detection Evasion (ICCV 2025 · similarity 0.91)
- 3D Gaussian Splatting Driven Multi-View Robust Physical Adversarial Camouflage Generation (ICCV 2025 · similarity 0.89)
- PhyCamo: A Robust Physical Camouflage via Contrastive Learning for Multi-View Physical Adversarial Attack (AAAI 2025 · similarity 0.87)
- RAUCA: A Novel Physical Adversarial Attack on Vehicle Detectors via Robust and Accurate Camouflage Generation (ICML 2024 · similarity 0.86)
- DepthVanish: Optimizing Adversarial Interval Structures for Stereo-Depth-Invisible Patches (NeurIPS 2025 · similarity 0.85)
- CNCA: Toward Customizable and Natural Generation of Adversarial Camouflage for Vehicle Detectors (NeurIPS 2024 · 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.
