Improving robustness against common corruptions by covariate shift adaptation
Today's state-of-the-art machine vision models are vulnerable to image corruptions like blurring or compression artefacts, limiting their performance in many real-world applications. We here argue that popular benchmarks to measure model robustness against common corruptions (like ImageNet-C) underestimate model robustness in many (but not all) application scenarios. The key insight is that in many scenarios, multiple unlabeled examples of the corruptions are available and can be used for unsupervised online adaptation. Replacing the activation statistics estimated by batch normalization on the training set with the statistics of the corrupted images consistently improves the robustness across 25 different popular computer vision models. Using the corrected statistics, ResNet-50 reaches 62.2% mCE on ImageNet-C compared to 76.7% without adaptation. With the more robust DeepAugment+AugMix model, we improve the state of the art achieved by a ResNet50 model up to date from 53.6% mCE to 45.4% mCE. Even adapting to a single sample improves robustness for the ResNet-50 and AugMix models, and 32 samples are sufficient to improve the current state of the art for a ResNet-50 architecture. We argue that results with adapted statistics should be included whenever reporting scores in corruption benchmarks and other out-of-distribution generalization settings.
Code (2)
Tasks
Domain GeneralizationImage ClassificationOut-of-Distribution GeneralizationUnsupervised Domain AdaptationMethods 이 논문이 사용한 방법론
Similar Papers 제목 키워드 기반
Tackling covariate shift with node-based Bayesian neural networks
Bayesian neural networks (BNNs) promise improved generalization under covariate shift by providing principled probabilistic representations of epistemic uncertainty. However, weight-based BNNs often struggle with high co…
Diversityimage-classificationImage ClassificationDiverse Gaussian Noise Consistency Regularization for Robustness and Uncertainty Calibration
Deep neural networks achieve high prediction accuracy when the train and test distributions coincide. In practice though, various types of corruptions occur which deviate from this setup and cause severe performance degr…
Data AugmentationImage ClassificationBenchmarking the Robustness of UAV Tracking Against Common Corruptions
The robustness of unmanned aerial vehicle (UAV) tracking is crucial in many tasks like surveillance and robotics. Despite its importance, little attention is paid to the performance of UAV trackers under common corruptio…
BenchmarkingTrimmed Maximum Likelihood Estimation for Robust Learning in Generalized Linear Models
We study the problem of learning generalized linear models under adversarial corruptions. We analyze a classical heuristic called the iterative trimmed maximum likelihood estimator which is known to be effective against …
regressionFourier-Based Augmentations for Improved Robustness and Uncertainty Calibration
Diverse data augmentation strategies are a natural approach to improving robustness in computer vision models against unforeseen shifts in data distribution. However, the ability to tailor such strategies to inoculate a …
Data AugmentationImage Augmentation