mlsanityData to 7 Oct 2026

A Fourier Perspective on Model Robustness in Computer Vision

arXiv 1906.08988 ↗First posted 21 Jun 2019cs.LG cs.CV stat.MLcited by 288

Not on the map. Why?

Abstract (arXiv)

Achieving robustness to distributional shift is a longstanding and challenging goal of computer vision. Data augmentation is a commonly used approach for improving robustness, however robustness gains are typically not uniform across corruption types. Indeed increasing performance in the presence of random noise is often met with reduced performance on other corruptions such as contrast change. Understanding when and why these sorts of trade-offs occur is a crucial step towards mitigating them. Towards this end, we investigate recently observed trade-offs caused by Gaussian data augmentation and adversarial training. We find that both methods improve robustness to corruptions that are concentrated in the high frequency domain while reducing robustness to corruptions that are concentrated in the low frequency domain. This suggests that one way to mitigate these trade-offs via data augmentation is to use a more diverse set of augmentations. Towards this end we observe that AutoAugment, a recently proposed data augmentation policy optimized for clean accuracy, achieves state-of-the-art robustness on the CIFAR-10-C benchmark.

Key facts. Cited by 288 papers in the map (data through 7 Oct 2026). First posted on arXiv 21 Jun 2019.

Builds on

No reference list yet.

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2021: 62; 2022: 53; 2023: 50; 2024: 47; 2025: 41; 2026: 35.

288 papers.

Sort:

By area:

  1. 2022Continual Test-Time Domain Adaptationcited by 520
  2. 2021MEMO: Test Time Robustness via Adaptation and Augmentationcited by 376
  3. 2022How Do Vision Transformers Work?cited by 305
  4. 2021Vision Transformers are Robust Learnerscited by 208
  5. 2023FreeU: Free Lunch in Diffusion U-Netcited by 195
  6. 2021PixMix: Dreamlike Pictures Comprehensively Improve Safety Measurescited by 131
  7. 2022Frequency Domain Model Augmentation for Adversarial Attackcited by 128
  8. 2022Inception Transformercited by 99
  9. 2021On Interaction Between Augmentations and Corruptions in Natural Corruption Robustnesscited by 98
  10. 2023A Comprehensive Study on Robustness of Image Classification Models: Benchmarking and Rethinkingcited by 85
  11. 2022Frequency-driven Imperceptible Adversarial Attack on Semantic Similaritycited by 76
  12. 2022Improving Vision Transformers by Revisiting High-frequency Componentscited by 75
38 more
  1. 2021On the effectiveness of adversarial training against common corruptionscited by 71
  2. 2022Benchmarking and Analyzing Point Cloud Classification under Corruptionscited by 70
  3. 2021Editing a classifier by rewriting its prediction rulescited by 64
  4. 2023On the Foundations of Shortcut Learningcited by 64
  5. 2021A Winning Hand: Compressing Deep Networks Can Improve Out-Of-Distribution Robustnesscited by 62
  6. 2021Adversarial Robustness Comparison of Vision Transformer and MLP-Mixer to CNNscited by 57
  7. 2021Causally motivated Shortcut Removal Using Auxiliary Labelscited by 55
  8. 2023Open-Vocabulary Segmentation with Semantic-Assisted Calibrationcited by 49
  9. 2023Spectrum-guided Multi-granularity Referring Video Object Segmentationcited by 48
  10. 2024Frequency-aware Feature Fusion for Dense Image Predictioncited by 48
  11. 2023Adaptive Frequency Filters As Efficient Global Token Mixerscited by 45
  12. 2022EfficientTrain: Exploring Generalized Curriculum Learning for Training Visual Backbonescited by 42
  13. 2021When Human Pose Estimation Meets Robustness: Adversarial Algorithms and Benchmarkscited by 41
  14. 2022Deep Frequency Filtering for Domain Generalizationcited by 41
  15. 2021Fighting deepfakes by detecting GAN DCT anomaliescited by 39
  16. 2021Pyramid Adversarial Training Improves ViT Performancecited by 39
  17. 2022PASTA: Proportional Amplitude Spectrum Training Augmentation for Syn-to-Real Domain Generalizationcited by 38
  18. 2021Spectral Bias in Practice: The Role of Function Frequency in Generalizationcited by 36
  19. 2022Exposing Outlier Exposure: What Can Be Learned From Few, One, and Zero Outlier Imagescited by 35
  20. 2023Continual Test-time Domain Adaptation via Dynamic Sample Selectioncited by 35
  21. 2021Improving robustness against common corruptions with frequency biased modelscited by 34
  22. 2022Do better ImageNet classifiers assess perceptual similarity better?cited by 34
  23. 2024Frequency-Adaptive Dilated Convolution for Semantic Segmentationcited by 34
  24. 2021Does enhanced shape bias improve neural network robustness to common corruptions?cited by 32
  25. 2021SpectralDefense: Detecting Adversarial Attacks on CNNs in the Fourier Domaincited by 30
  26. 2022Large-scale Robustness Analysis of Video Action Recognition Modelscited by 29
  27. 2021Defending Against Image Corruptions Through Adversarial Augmentationscited by 28
  28. 2021A Frequency Perspective of Adversarial Robustnesscited by 28
  29. 2023What do neural networks learn in image classification? A frequency shortcut perspectivecited by 28
  30. 2023Spatial-frequency channels, shape bias, and adversarial robustnesscited by 28
  31. 2021Enhancing Adversarial Robustness via Test-time Transformation Ensemblingcited by 27
  32. 2021Impact of Aliasing on Generalization in Deep Convolutional Networkscited by 25
  33. 2022Robustness Analysis of Video-Language Models Against Visual and Language Perturbationscited by 23
  34. 2022On the Strong Correlation Between Model Invariance and Generalizationcited by 23
  35. 2022Global Spectral Filter Memory Network for Video Object Segmentationcited by 22
  36. 2024Adversarial Attacks of Vision Tasks in the Past 10 Years: A Surveycited by 22
  37. 2023A Survey on the Robustness of Computer Vision Models against Common Corruptionscited by 20
  38. 2024Can Biases in ImageNet Models Explain Generalization?cited by 20

50 most cited of 288 All 288 (JSON)