Research map / Adversarial robustness and security / Attacks on text and graphs
Gradient-Reweighted Adversarial Camouflage for Physical Object Detection Evasion
Jiawei Liang, Siyuan Liang, Tianrui Lou, Ming Zhang, Wenjin Li, Dunqiu Fan, 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 gradient reweighting (ablation) | GRAC baseline texture optimization · YOLOv3 white-box | AP@0.5: 19.02 -> 15.56 | Improved |
| add illumination simulation (ablation) | GRAC baseline · YOLOv3 white-box | AP@0.5: 19.02 -> 16.48 | Improved |
| add texture regularization (pooling) (ablation) | GRAC baseline · YOLOv3 white-box | AP@0.5: 19.02 -> 11.86 | Improved |
| add all three components (ablation) | GRAC baseline · YOLOv3 white-box | AP@0.5: 19.02 -> 2.94 | Improved |
| add gradient reweighting (ablation) | GRAC baseline · Faster R-CNN black-box | AP@0.5: 35.14 -> 26.85 | Improved |
| add illumination simulation (ablation) | GRAC baseline · Faster R-CNN black-box | AP@0.5: 35.14 -> 30.39 | Improved |
| add texture regularization (ablation) | GRAC baseline · Faster R-CNN black-box | AP@0.5: 35.14 -> 28.18 | Improved |
| add all three components (ablation) | GRAC baseline · PVT black-box | AP@0.5: 23.39 -> 18.03 | Improved |
| replace best prior method RAUCA with GRAC | adversarial camouflage · YOLOv3 elevation average | AP@0.5: 16.31 -> 2.94 | Improved |
| replace RAUCA with GRAC on black-box detectors | adversarial camouflage · Mask R-CNN | AP@0.5: 28.90 -> 29.31 | No clear effect |
| increase gradient accumulation steps (ablation) | gradient reweighting · YOLOv3 white-box vs black-box | AP@0.5 | Mixed |
| increase texture pooling size (ablation) | texture regularization · pooling-size sweep | AP@0.5 | Mixed |
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 8 of 9 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
- Deformable and Multi-view Gradient-Aligned Physical Adversarial Camouflage (ECCV 2026 · similarity 0.91)
- ACTIVE: Towards Highly Transferable 3D Physical Camouflage for Universal and Robust Vehicle Evasion (ICCV 2023 · similarity 0.90)
- RAUCA: A Novel Physical Adversarial Attack on Vehicle Detectors via Robust and Accurate Camouflage Generation (ICML 2024 · similarity 0.90)
- CNCA: Toward Customizable and Natural Generation of Adversarial Camouflage for Vehicle Detectors (NeurIPS 2024 · similarity 0.89)
- PhyCamo: A Robust Physical Camouflage via Contrastive Learning for Multi-View Physical Adversarial Attack (AAAI 2025 · similarity 0.88)
- Adversarial Attacks on Event-Based Pedestrian Detectors: A Physical Approach (AAAI 2025 · similarity 0.86)
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.
