CutMix: Regularization Strategy to Train Strong Classifiers with Localizable Features

TL;DR

CutMix is a novel data augmentation technique that improves CNN performance by combining images and labels, leading to better generalization, localization, robustness, and transferability across multiple vision tasks.

Contribution

This paper introduces CutMix, a new augmentation method that mixes image patches and labels, outperforming existing regional dropout strategies in classification, localization, detection, and captioning tasks.

Findings

Outperforms state-of-the-art augmentation methods on CIFAR and ImageNet.

Enhances weakly-supervised localization and transfer learning performance.

Improves model robustness against input corruptions.

Abstract

Regional dropout strategies have been proposed to enhance the performance of convolutional neural network classifiers. They have proved to be effective for guiding the model to attend on less discriminative parts of objects (e.g. leg as opposed to head of a person), thereby letting the network generalize better and have better object localization capabilities. On the other hand, current methods for regional dropout remove informative pixels on training images by overlaying a patch of either black pixels or random noise. Such removal is not desirable because it leads to information loss and inefficiency during training. We therefore propose the CutMix augmentation strategy: patches are cut and pasted among training images where the ground truth labels are also mixed proportionally to the area of the patches. By making efficient use of training pixels and retaining the regularization effect of regional dropout, CutMix consistently outperforms the state-of-the-art augmentation strategies on CIFAR and ImageNet classification tasks, as well as on the ImageNet weakly-supervised localization task. Moreover, unlike previous augmentation methods, our CutMix-trained ImageNet classifier, when used as a pretrained model, results in consistent performance gains in Pascal detection and MS-COCO image captioning benchmarks. We also show that CutMix improves the model robustness against input corruptions and its out-of-distribution detection performances. Source code and pretrained models are available at https://github.com/clovaai/CutMix-PyTorch .