Data to 7 Oct 2026

Universal Adversarial Perturbations Against Semantic Image Segmentation

arXiv 1704.05712 ↗First posted 19 Apr 2017stat.ML cs.AI cs.CVcited by 92

Not on the map. Why?

Abstract (arXiv)

While deep learning is remarkably successful on perceptual tasks, it was also shown to be vulnerable to adversarial perturbations of the input. These perturbations denote noise added to the input that was generated specifically to fool the system while being quasi-imperceptible for humans. More severely, there even exist universal perturbations that are input-agnostic but fool the network on the majority of inputs. While recent work has focused on image classification, this work proposes attacks against semantic image segmentation: we present an approach for generating (universal) adversarial perturbations that make the network yield a desired target segmentation as output. We show empirically that there exist barely perceptible universal noise patterns which result in nearly the same predicted segmentation for arbitrary inputs. Furthermore, we also show the existence of universal noise which removes a target class (e.g., all pedestrians) from the segmentation while leaving the segmentation mostly unchanged otherwise.

Key facts. Cited by 92 papers in the map (data through 7 Oct 2026). First posted on arXiv 19 Apr 2017.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 27; 2022: 21; 2023: 21; 2024: 15; 2025: 4; 2026: 4.

92 papers.

Sort:

By area:

2023DDP: Diffusion Model for Dense Visual Predictioncited by 148
2022SegPGD: An Effective and Efficient Adversarial Attack for Evaluating and Boosting Segmentation Robustnesscited by 74
2021A Survey On Universal Adversarial Attackcited by 59
2021Evaluating the Robustness of Semantic Segmentation for Autonomous Driving against Real-World Adversarial Patch Attackscited by 55
2021CMUA-Watermark: A Cross-Model Universal Adversarial Watermark for Combating Deepfakescited by 54
2023A Survey on Transferability of Adversarial Examples across Deep Neural Networkscited by 43
2021CausalAdv: Adversarial Robustness through the Lens of Causalitycited by 37
2021Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safetycited by 35
2024Test-Time Backdoor Attacks on Multimodal Large Language Modelscited by 35
2024DarkSAM: Fooling Segment Anything Model to Segment Nothingcited by 32
2022On the Real-World Adversarial Robustness of Real-Time Semantic Segmentation Models for Autonomous Drivingcited by 30
2023Semantic Image Segmentation: Two Decades of Researchcited by 26

38 more

2021TAFIM: Targeted Adversarial Attacks against Facial Image Manipulationscited by 25
2021Evaluating Adversarial Attacks on Driving Safety in Vision-Based Autonomous Vehiclescited by 24
2022A Survey on Physical Adversarial Attack in Computer Visioncited by 24
2021Meta Adversarial Training against Universal Patchescited by 19
2022Watermark Vaccine: Adversarial Attacks to Prevent Watermark Removalcited by 18
2021Stereoscopic Universal Perturbations across Different Architectures and Datasetscited by 17
2021Towards Understanding Adversarial Robustness of Optical Flow Networkscited by 15
2021Adversarial Examples on Segmentation Models Can be Easy to Transfercited by 15
2024Adversarial Safety-Critical Scenario Generation using Naturalistic Human Driving Priorscited by 15
2021Universal Spectral Adversarial Attacks for Deformable Shapescited by 14
2024PB-UAP: Hybrid Universal Adversarial Attack For Image Segmentationcited by 14
2021The Vulnerability of Semantic Segmentation Networks to Adversarial Attacks in Autonomous Driving: Enhancing Extensive Environment Sensingcited by 12
2023Towards Reliable Evaluation and Fast Training of Robust Semantic Segmentation Modelscited by 12
2024Visual Robustness Benchmark for Visual Question Answering (VQA)cited by 12
2022DIMBA: Discretely Masked Black-Box Attack in Single Object Trackingcited by 11
2021When and How to Fool Explainable Models (and Humans) with Adversarial Examplescited by 10
2022Defending From Physically-Realizable Adversarial Attacks Through Internal Over-Activation Analysiscited by 10
2023Deep Learning Safety Concerns in Automated Driving Perceptioncited by 10
2023Targeted Adversarial Attacks on Generalizable Neural Radiance Fieldscited by 10
2023Adversarial Examples in the Physical World: A Surveycited by 10
2023TranSegPGD: Improving Transferability of Adversarial Examples on Semantic Segmentationcited by 10
2023Uncertainty-weighted Loss Functions for Improved Adversarial Attacks on Semantic Segmentationcited by 9
2021What Do Deep Nets Learn? Class-wise Patterns Revealed in the Input Spacecited by 8
2021Semantically Stealthy Adversarial Attacks against Segmentation Modelscited by 7
2021IPatch: A Remote Adversarial Patchcited by 7
2021UniNet: A Unified Scene Understanding Network and Exploring Multi-Task Relationships through the Lens of Adversarial Attackscited by 7
2022Runtime Monitoring for Out-of-Distribution Detection in Object Detection Neural Networkscited by 7
2021Selection of Source Images Heavily Influences the Effectiveness of Adversarial Attackscited by 6
2022Physical Passive Patch Adversarial Attacks on Visual Odometry Systemscited by 6
2022Adversarial Attacks on Monocular Pose Estimationcited by 6
2024Unsegment Anything by Simulating Deformationcited by 6
2024Adversarial Robustness Guarantees for Quantum Classifierscited by 6
2022Smoothed Embeddings for Certified Few-Shot Learningcited by 5
2023Influencer Backdoor Attack on Semantic Segmentationcited by 5
2023A Certified Radius-Guided Attack Framework to Image Segmentation Modelscited by 5
2023Uncertainty-based Detection of Adversarial Attacks in Semantic Segmentationcited by 5
2023A Review of Adversarial Attacks in Computer Visioncited by 5
2023Segment (Almost) Nothing: Prompt-Agnostic Adversarial Attacks on Segmentation Modelscited by 5

50 most cited of 92 All 92 (JSON)