mlsanityData to 14 Oct 2026

A study of the effect of JPG compression on adversarial images

arXiv 1608.00853 ↗First posted 2 Aug 2016cs.CV cs.LGcited by 197

Not on the map. Why?

Abstract (arXiv)

Neural network image classifiers are known to be vulnerable to adversarial images, i.e., natural images which have been modified by an adversarial perturbation specifically designed to be imperceptible to humans yet fool the classifier. Not only can adversarial images be generated easily, but these images will often be adversarial for networks trained on disjoint subsets of data or with different architectures. Adversarial images represent a potential security risk as well as a serious machine learning challenge---it is clear that vulnerable neural networks perceive images very differently from humans. Noting that virtually every image classification data set is composed of JPG images, we evaluate the effect of JPG compression on the classification of adversarial images. For Fast-Gradient-Sign perturbations of small magnitude, we found that JPG compression often reverses the drop in classification accuracy to a large extent, but not always. As the magnitude of the perturbations increases, JPG recompression alone is insufficient to reverse the effect.

Key facts. Cited by 197 papers in the map (data through 14 Oct 2026). First posted on arXiv 2 Aug 2016.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2021: 56; 2022: 20; 2023: 35; 2024: 34; 2025: 35; 2026: 17.

197 papers.

Sort:

By area:

  1. 2024Agent Smith: A Single Image Can Jailbreak One Million Multimodal LLM Agents Exponentially Fastcited by 153
  2. 2022Frequency Domain Model Augmentation for Adversarial Attackcited by 130
  3. 2021Invisible for both Camera and LiDAR: Security of Multi-Sensor Fusion based Perception in Autonomous Driving Under Physical-World Attackscited by 101
  4. 2024Exploring the Adversarial Vulnerabilities of Vision-Language-Action Models in Roboticscited by 73
  5. 2021SoK: How Robust is Image Classification Deep Neural Network Watermarking? (Extended Version)cited by 69
  6. 2021Segment and Complete: Defending Object Detectors against Adversarial Patch Attacks with Robust Patch Detectioncited by 67
  7. 2023How Deep Learning Sees the World: A Survey on Adversarial Attacks & Defensescited by 66
  8. 2022Physical Attack on Monocular Depth Estimation with Optimal Adversarial Patchescited by 57
  9. 2021AdvDrop: Adversarial Attack to DNNs by Dropping Informationcited by 56
  10. 2021Person Re-identification Method Based on Color Attack and Joint Defencecited by 56
  11. 2021Meta Gradient Adversarial Attackcited by 55
  12. 2021Exploring Adversarial Robustness of Multi-Sensor Perception Systems in Self Drivingcited by 45
38 more
  1. 2021Reveal of Vision Transformers Robustness against Adversarial Attackscited by 43
  2. 2021A Comprehensive Evaluation Framework for Deep Model Robustnesscited by 42
  3. 2022Sparsity Winning Twice: Better Robust Generalization from More Efficient Trainingcited by 41
  4. 2024One Perturbation is Enough: On Generating Universal Adversarial Perturbations against Vision-Language Pre-training Modelscited by 40
  5. 2023Image Shortcut Squeezing: Countering Perturbative Availability Poisons with Compressioncited by 39
  6. 2024Copyright Protection in Generative AI: A Technical Perspectivecited by 38
  7. 2024Efficient Generation of Targeted and Transferable Adversarial Examples for Vision-Language Models Via Diffusion Modelscited by 38
  8. 2023DiffProtect: Generate Adversarial Examples with Diffusion Models for Facial Privacy Protectioncited by 35
  9. 2022DISCO: Adversarial Defense with Local Implicit Functionscited by 34
  10. 2023Robustness of SAM: Segment Anything Under Corruptions and Beyondcited by 31
  11. 2022PatchZero: Defending against Adversarial Patch Attacks by Detecting and Zeroing the Patchcited by 30
  12. 2021JEM++: Improved Techniques for Training JEMcited by 28
  13. 2021A Frequency Perspective of Adversarial Robustnesscited by 28
  14. 2024Revisiting the Adversarial Robustness of Vision Language Models: a Multimodal Perspectivecited by 28
  15. 2023EditShield: Protecting Unauthorized Image Editing by Instruction-guided Diffusion Modelscited by 27
  16. 2021Towards Robust Neural Image Compression: Adversarial Attack and Model Finetuningcited by 25
  17. 2022A Comprehensive Study of the Robustness for LiDAR-based 3D Object Detectors against Adversarial Attackscited by 22
  18. 2023TPatch: A Triggered Physical Adversarial Patchcited by 22
  19. 2021Multi-granularity Textual Adversarial Attack with Behavior Cloningcited by 21
  20. 2021Generative Dynamic Patch Attackcited by 19
  21. 2024CLIP-Guided Generative Networks for Transferable Targeted Adversarial Attackscited by 19
  22. 2024StegoGAN: Leveraging Steganography for Non-Bijective Image-to-Image Translationcited by 18
  23. 2021MagDR: Mask-guided Detection and Reconstruction for Defending Deepfakescited by 17
  24. 2023Deep Intellectual Property Protection: A Surveycited by 17
  25. 2022Towards Bridging the Performance Gaps of Joint Energy-based Modelscited by 15
  26. 2023DIFFender: Diffusion-Based Adversarial Defense against Patch Attackscited by 15
  27. 2021Adversarial Attacks on Multi-task Visual Perception for Autonomous Drivingcited by 14
  28. 2021A Person Re-identification Data Augmentation Method with Adversarial Defense Effectcited by 13
  29. 2023Black-box Targeted Adversarial Attack on Segment Anything (SAM)cited by 13
  30. 2024Object Detectors in the Open Environment: Challenges, Solutions, and Outlookcited by 13
  31. 2021Learning Diverse-Structured Networks for Adversarial Robustnesscited by 12
  32. 2021Staircase Sign Method for Boosting Adversarial Attackscited by 12
  33. 2023AdvART: Adversarial Art for Camouflaged Object Detection Attackscited by 12
  34. 2023SAAM: Stealthy Adversarial Attack on Monocular Depth Estimationcited by 12
  35. 2021Multi-objective Search of Robust Neural Architectures against Multiple Types of Adversarial Attackscited by 11
  36. 2021Adversarial Feature Augmentation and Normalization for Visual Recognitioncited by 11
  37. 2022Robust Real-World Image Super-Resolution against Adversarial Attackscited by 11
  38. 2022An Evolutionary, Gradient-Free, Query-Efficient, Black-Box Algorithm for Generating Adversarial Instances in Deep Networkscited by 11

50 most cited of 197 All 197 (JSON)