ECCV 2026 accepted papers on Adversarial robustness and security
39 of the 2,834 papers accepted at ECCV 2026 (1.4%) are on Adversarial robustness and security, compared with 2.6% at ECCV 2024. They fall into 4 topics.
See ECCV 2026 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Attacks on text and graphs · 20 papers
- VLA-Hijack: A Transferable Patch Attack against Vision-Language-Action Models via Visual Proprioception Hijacking (7 citations)
- Universal Image Immunization against Diffusion-based Image Editing via Semantic Injection (6 citations)
- Read or Ignore? A Unified Benchmark for Typographic-Attack Robustness and Text Recognition in Vision-Language Models (5 citations)
- Contextual Image Attack: How Visual Context Exposes Multimodal Safety Vulnerabilities (3 citations)
- Fast and Flexible Robustness Certificates for Semantic Segmentation (2 citations)
- Rethinking Robust Adversarial Concept Erasure in Diffusion Models (2 citations)
- Are GUI Agents Focused Enough? Automated Distraction via Semantic-level UI Element Injection (1 citation)
- Robust onion: Peeling Open Vocab Object Detectors Under Noise (1 citation)
- Adversarial Attack and Disturbance Detection by Hadamard-Coded Output Representations for Object Detection and Semantic Segmentation
- An Inverse-Adversarial and Difficulty-Adaptive Robust Vision-Language Model
- Breaking Rigidity in Adversarial Patch Attacks
- Deformable and Multi-view Gradient-Aligned Physical Adversarial Camouflage
and 8 more in the interactive map.
Black-box and transfer attacks · 7 papers
- Scaling Laws for Black-box Adversarial Attacks (11 citations)
- Stealthy Multi-task Adversarial Attacks (3 citations)
- A Classifier-Agnostic Zero-Shot Adversarial Attack Detection via CLIP
- Defending from GeoLocalization through Adversarial Road Trips
- Learning with Bilevel-Minimax Optimization for Efficient and Reliable Transfer Attacks
- Mechanistic interventions for explainable digital pathology uncovers adversarial vulnerabilities
- SegPAR: Class-Centric Decision-Based Sparse Attack for Semantic Segmentation
Backdoor and poisoning · 6 papers
- Breaking High Confidence: Practical Face Impersonation under High-Security Thresholds
- Capacity Overflow: A Blind Spot for Backdoor Attacks in Vision MoE
- Indelible Backdoors: On the Limits of Post-Training Defenses
- On the Vulnerability of Parameter-Level Defenses to Model Merging
- Rethink Backdoor Robustness in Vision Transformers
- TooBad: Backdoor Diffusion Models with Ultra-Low Poison Rate and Imperceptible Trigger
Certified and empirical robustness · 6 papers
- Don’t Teach Instability, Teach Robustness: Selective Sensitivity Gating for Adversarial Robust Distillation
- GCMRD: Global Consistency Multi-teacher Robustness Distillation
- Improving Adversarial Robustness by Mitigating Instability through Relearning
- Improving Adversarial Robustness via Activation Amplification and Attenuation
- Robustness Meets Uncertainty: Evidential Adversarial Training for Robust Selective Classification
- RoME: Robust Mixture of Low-Rank Experts against Multiple Adversarial Perturbations
