mlsanityData to 7 Oct 2026

Towards Robust Neural Image Compression: Adversarial Attack and Model Finetuning

arXiv 2112.08691 ↗First posted 16 Dec 2021cs.CV cs.MM eess.IVcited by 24

Not on the map. Why?

Abstract (arXiv)

Deep neural network-based image compression has been extensively studied. However, the model robustness which is crucial to practical application is largely overlooked. We propose to examine the robustness of prevailing learned image compression models by injecting negligible adversarial perturbation into the original source image. Severe distortion in decoded reconstruction reveals the general vulnerability in existing methods regardless of their settings (e.g., network architecture, loss function, quality scale). A variety of defense strategies including geometric self-ensemble based pre-processing, and adversarial training, are investigated against the adversarial attack to improve the model's robustness. Later the defense efficiency is further exemplified in real-life image recompression case studies. Overall, our methodology is simple, effective, and generalizable, making it attractive for developing robust learned image compression solutions. All materials are made publicly accessible at https://njuvision.github.io/RobustNIC for reproducible research.

Key facts. Cited by 24 papers in the map (data through 7 Oct 2026). Builds on 45 of its 78 references that are in the map. First posted on arXiv 16 Dec 2021.

Builds on

45 of 78 references.

References · 45

  1. 2015Deep Residual Learning for Image Recognitioncited by 51,809
  2. 2014Adam: A Method for Stochastic Optimizationcited by 36,960
  3. 2018The Unreasonable Effectiveness of Deep Features as a Perceptual Metriccited by 11,029
  4. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  5. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
  6. 2013Intriguing properties of neural networkscited by 5,114
  7. 2016The Cityscapes Dataset for Semantic Urban Scene Understandingcited by 4,427
  8. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,976
  9. 2016Adversarial examples in the physical worldcited by 1,762
  10. 2017Enhanced Deep Residual Networks for Single Image Super-Resolutioncited by 1,543
  11. 2015DeepFool: a simple and accurate method to fool deep neural networkscited by 1,315
  12. 2018Variational image compression with a scale hyperpriorcited by 910
33 more
  1. 2017Ensemble Adversarial Training: Attacks and Defensescited by 833
  2. 2014Deep Neural Networks are Easily Fooled: High Confidence Predictions for Unrecognizable Imagescited by 774
  3. 2016End-to-end Optimized Image Compressioncited by 673
  4. 2018Joint Autoregressive and Hierarchical Priors for Learned Image Compressioncited by 642
  5. 2017Video Enhancement with Task-Oriented Flowcited by 612
  6. 2020Learned Image Compression with Discretized Gaussian Mixture Likelihoods and Attention Modulescited by 570
  7. 2017Feature Squeezing: Detecting Adversarial Examples in Deep Neural Networkscited by 484
  8. 2015Autoencoding beyond pixels using a learned similarity metriccited by 407
  9. 2017Mitigating Adversarial Effects Through Randomizationcited by 382
  10. 2020High-Fidelity Generative Image Compressioncited by 331
  11. 2020CompressAI: a PyTorch library and evaluation platform for end-to-end compression researchcited by 271
  12. 2016A study of the effect of JPG compression on adversarial imagescited by 196
  13. 2018Context-adaptive Entropy Model for End-to-end Optimized Image Compressioncited by 138
  14. 2015Density Modeling of Images using a Generalized Normalization Transformationcited by 123
  15. 2018Transferable Adversarial Attacks for Image and Video Object Detectioncited by 109
  16. 2021Enhanced Invertible Encoding for Learned Image Compressioncited by 102
  17. 2019Non-local Attention Optimized Deep Image Compressioncited by 86
  18. 2019Image and Video Compression with Neural Networks: A Reviewcited by 82
  19. 2021Improving Adversarial Robustness via Channel-wise Activation Suppressingcited by 76
  20. 2017Learning Convolutional Networks for Content-weighted Image Compressioncited by 73
  21. 2015Learning with a Strong Adversarycited by 68
  22. 2021Transformer-based Image Compressioncited by 60
  23. 2017Adversarial examples for generative modelscited by 58
  24. 2020Learning End-to-End Lossy Image Compression: A Benchmarkcited by 56
  25. 2019FDA: Feature Disruptive Attackcited by 55
  26. 2019Attacking Optical Flowcited by 41
  27. 2018Characterizing Adversarial Examples Based on Spatial Consistency Information for Semantic Segmentationcited by 40
  28. 2019Evaluating Robustness of Deep Image Super-Resolution against Adversarial Attackscited by 37
  29. 2020Lossy Image Compression with Normalizing Flowscited by 25
  30. 2016Adversarial Images for Variational Autoencoderscited by 19
  31. 2021Deep Image Destruction: Vulnerability of Deep Image-to-Image Models against Adversarial Attackscited by 8
  32. 2019Image Transformation can make Neural Networks more robust against Adversarial Examplescited by 3
  33. 2019Extreme Image Coding via Multiscale Autoencoders With Generative Adversarial Optimizationcited by 3

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2022: 3; 2023: 3; 2024: 10; 2025: 4; 2026: 4.

24 papers.

Sort:

By area:

  1. 2023MLIC++: Linear Complexity Multi-Reference Entropy Modeling for Learned Image Compressioncited by 56
  2. 2024Revisiting the Adversarial Robustness of Vision Language Models: a Multimodal Perspectivecited by 28
  3. 2022Pre-trained Adversarial Perturbationscited by 16
  4. 2025Align is not Enough: Multimodal Universal Jailbreak Attack against Multimodal Large Language Modelscited by 8
  5. 2024A Training-Free Defense Framework for Robust Learned Image Compressioncited by 5
  6. 2022MALICE: Manipulation Attacks on Learned Image ComprEssioncited by 3
  7. 2023Improving Transferability of Adversarial Examples via Bayesian Attackscited by 3
  8. 2024A Taxonomy of Miscompressions: Preparing Image Forensics for Neural Compressioncited by 3
  9. 2024Exploring adversarial robustness of JPEG AI: methodology, comparison and new methodscited by 3
  10. 2024Transferable Learned Image Compression-Resistant Adversarial Perturbationscited by 2
  11. 2024Survey on Visual Signal Coding and Processing with Generative Models: Technologies, Standards and Optimizationcited by 2
  12. 2024Approximately Invertible Neural Network for Learned Image Compressioncited by 2
12 more
  1. 2024Take Fake as Real: Realistic-like Robust Black-box Adversarial Attack to Evade AIGC Detectioncited by 2
  2. 2022Improving Multi-generation Robustness of Learned Image Compressioncited by 1
  3. 2023Spatial-Frequency Discriminability for Revealing Adversarial Perturbationscited by 1
  4. 2024SLIC: Secure Learned Image Codec through Compressed Domain Watermarking to Defend Image Manipulationcited by 1
  5. 2025Bitstream Collisions in Neural Image Compression via Adversarial Perturbationscited by 1
  6. 2025NIC-RobustBench: A Comprehensive Open-Source Toolkit for Neural Image Compression and Robustness Analysiscited by 1
  7. 2025T-MLA: A targeted multiscale log-exponential attack framework for neural image compressioncited by 1
  8. 2026Explanation-Guided Adversarial Training for Robust and Interpretable Modelscited by 1
  9. 2026Uni-LVC: A Unified Method for Intra- and Inter-Mode Learned Video Compressioncited by 1
  10. 2024Universal Multi-view Black-box Attack against Object Detectors via Layout Optimizationcited by 0
  11. 2026Control Your View: High-Resolution Global Semantic Manipulation in Learned Image Compressioncited by 0
  12. 2026Fast Test-Time Refinement for Robust Learned Image Compressioncited by 0