Research map / 3D vision / Gaussian splatting
3D Gaussian Splatting Driven Multi-View Robust Physical Adversarial Camouflage Generation
Tianrui Lou, Xiaojun Jia, Siyuan Liang, Jiawei Liang, Ming Zhang, Yanjun Xiao, 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 multi-view camouflage consistency (ablation) | PGA · CARLA digital ablation | AP@0.5 average (lower is better): 27.32 -> 22.87 | Improved |
| add min-max optimization with background counter-attack (ablation) | PGA · CARLA digital ablation | AP@0.5 average (lower is better): 30.50 -> 22.87 | Improved |
| add multi-view consistency (ablation) | PGA · CARLA digital, Faster R-CNN | AP@0.5: 8.05 -> 3.57 | Improved |
| add min-max optimization (ablation) | PGA · CARLA digital, Faster R-CNN | AP@0.5: 10.23 -> 3.57 | Improved |
| PGA instead of RAUCA at 5m | PGA · CARLA digital, 5m | AP@0.5 average: 37.16 -> 23.46 | Improved |
| PGA instead of RAUCA at 10m | PGA · CARLA digital, 10m | AP@0.5 average: 39.25 -> 21.78 | Improved |
| PGA instead of RAUCA at 20m | PGA · CARLA digital, 20m | AP@0.5 average: 49.50 -> 20.73 | Improved |
| PGA instead of RAUCA across pitch angles | PGA · CARLA pitch sweep | AP@0.5 average: 33.97 -> 6.73 | Improved |
| PGA instead of RAUCA on 1:24 toy car at 50cm | PGA · 1:24 physical, 50cm | AP@0.5 average: 36.69 -> 28.69 | Improved |
| PGA instead of RAUCA on 1:24 toy car at 100cm | PGA · 1:24 physical, 100cm | AP@0.5 average: 37.25 -> 28.26 | Improved |
| deploy PGA camouflage on real vehicle | PGA · 1:1 real vehicle | AP@0.5 Faster R-CNN: 88.48 -> 25.67 | Improved |
| transfer PGA from Faster R-CNN to YOLO-v5 (5m sunny) | PGA · CARLA digital, 5m | AP@0.5: 4.52 -> 39.10 | Mixed |
Closest papers by meaning
- PhyCamo: A Robust Physical Camouflage via Contrastive Learning for Multi-View Physical Adversarial Attack (AAAI 2025 · similarity 0.90)
- Deformable and Multi-view Gradient-Aligned Physical Adversarial Camouflage (ECCV 2026 · similarity 0.89)
- Gradient-Reweighted Adversarial Camouflage for Physical Object Detection Evasion (ICCV 2025 · similarity 0.89)
- RAUCA: A Novel Physical Adversarial Attack on Vehicle Detectors via Robust and Accurate Camouflage Generation (ICML 2024 · similarity 0.87)
- SecureGS: Boosting the Security and Fidelity of 3D Gaussian Splatting Steganography (ICLR 2025 · similarity 0.86)
- Towards Transferable Targeted 3D Adversarial Attack in the Physical World (CVPR 2024 · similarity 0.85)
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.
