FFT-based Selection and Optimization of Statistics for Robust Recognition of Severely Corrupted Images
Not on the map. Why?
Abstract (arXiv)
Improving model robustness in case of corrupted images is among the key challenges to enable robust vision systems on smart devices, such as robotic agents. Particularly, robust test-time performance is imperative for most of the applications. This paper presents a novel approach to improve robustness of any classification model, especially on severely corrupted images. Our method (FROST) employs high-frequency features to detect input image corruption type, and select layer-wise feature normalization statistics. FROST provides the state-of-the-art results for different models and datasets, outperforming competitors on ImageNet-C by up to 37.1% relative gain, improving baseline of 40.9% mCE on severe corruptions.
Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Builds on 19 of its 26 references that are in the map. First posted on arXiv 21 Mar 2024.
Builds on
19 of 26 references.
References · 19
- 2015Deep Residual Learning for Image Recognitioncited by 51,809
- 2020An Image is Worth 16x16 Words: Transformers for Image Recognition at Scalecited by 29,915
- 2014ImageNet Large Scale Visual Recognition Challengecited by 8,441
- 2019Benchmarking Neural Network Robustness to Common Corruptions and Perturbationscited by 2,857
- 2020The Many Faces of Robustness: A Critical Analysis of Out-of-Distribution Generalizationcited by 1,842
- 2019AugMix: A Simple Data Processing Method to Improve Robustness and Uncertaintycited by 897
- 2019Making Convolutional Networks Shift-Invariant Againcited by 390
- 2021Are Transformers More Robust Than CNNs?cited by 187
- 2021PixMix: Dreamlike Pictures Comprehensively Improve Safety Measurescited by 131
- 2022Frequency Domain Model Augmentation for Adversarial Attackcited by 128
- 2019Improving Robustness Without Sacrificing Accuracy with Patch Gaussian Augmentationcited by 105
- 2023TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptationcited by 101
7 more
- 2021AugMax: Adversarial Composition of Random Augmentations for Robust Trainingcited by 89
- 2021Amplitude-Phase Recombination: Rethinking Robustness of Convolutional Neural Networks in Frequency Domaincited by 70
- 2021SITA: Single Image Test-time Adaptationcited by 57
- 2021Improving robustness against common corruptions with frequency biased modelscited by 34
- 2023HybridAugment++: Unified Frequency Spectra Perturbations for Model Robustnesscited by 12
- 2017Improving Robustness of Feature Representations to Image Deformations using Powered Convolution in CNNscited by 2
- 2020On Performance of Multiscale Sparse Fast Fourier Transform Algorithmcited by 1
Built on by
Citing papers per year: 2024: 1.
1 paper.