Research map / Detection and segmentation / CNN features
Context-Aware Transfer Attacks for Object Detection
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 |
|---|---|---|---|
| Use context-graph-selected helper labels and placements instead of victim-only attacks | Context-aware sequential attack · PASCAL VOC, blackbox detectors, all tested perturbation budgets | fooling rate: 5--20 percentage points | Helped |
| Use context-aware helper assignments instead of random helper labels | Context-aware sequential attack · PASCAL VOC, L∞ ≤ 20, blackbox RetinaNet | fooling rate: 35.2 vs 31.0 | Helped |
| Use context-aware helper assignments instead of random helper labels | Context-aware sequential attack · PASCAL VOC, L∞ ≤ 20, blackbox Libra R-CNN | fooling rate: 38.4 vs 29.6 | Helped |
| Use context-aware helper assignments instead of random helper labels | Context-aware sequential attack · MS COCO, L∞ ≤ 20, blackbox DETR | fooling rate: 20.0 vs 20.2 | Hurt |
| Add helper objects sequentially to the attack plan (ablation) | Context-aware sequential attack · PASCAL VOC and MS COCO | fooling rate | Helped |
| Increase helper count beyond the first few objects (ablation) | Context-aware sequential attack · PASCAL VOC and MS COCO | fooling rate | No clear effect |
| Choose helper labels and locations at random rather than using context information (ablation) | Context-aware sequential attack · PASCAL VOC and MS COCO | fooling rate | Hurt |
Closest papers by meaning
- Transferable Adversarial Attacks for Object Detection Using Object-Aware Significant Feature Distortion (AAAI 2024 · similarity 0.92)
- Amnesia as a Catalyst for Enhancing Black Box Pixel Attacks in Image Classification and Object Detection (NeurIPS 2024 · similarity 0.84)
- Everywhere Attack: Attacking Locally and Globally to Boost Targeted Transferability (AAAI 2025 · similarity 0.84)
- GLOW: Global Layout Aware Attacks on Object Detection (CVPR 2024 · similarity 0.84)
- Rethinking Model Ensemble in Transfer-based Adversarial Attacks (ICLR 2024 · similarity 0.84)
- ContextShift: A Controlled Benchmark for Context Dependence in Object Detection (NeurIPS 2026 · 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.