KeepAugment: A Simple Information-Preserving Data Augmentation Approach
Authors:
- 1 June 2021
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1055-1064
Abstract
Data augmentation (DA) is an essential technique for training state-of-the-art deep learning systems. In this paper, we empirically show that the standard data augmentation methods may introduce distribution shift and consequently hurt the performance on unaugmented data during inference. To alleviate this issue, we propose a simple yet effective approach, dubbed KeepAugment, to increase the fidelity of augmented images. The idea is to use the saliency map to detect important regions on the original images and preserve these informative regions during augmentation. This information-preserving strategy allows us to generate more faithful training examples. Empirically, we demonstrate that our method significantly improves upon a number of prior art data augmentation schemes, e.g. AutoAugment, Cutout, random erasing, achieving promising results on image classification, semi-supervised image classification, multi-view multi-camera tracking and object detection.Keywords
All Related Versions
This publication has 14 references indexed in Scilit:
- CutMix: Regularization Strategy to Train Strong Classifiers With Localizable Features2019
- Focal Loss for Dense Object Detection2017
- Grad-CAM: Visual Explanations from Deep Networks via Gradient-Based Localization2017
- Feature Pyramid Networks for Object Detection2017
- Deep Residual Learning for Image Recognition2016
- Scalable Person Re-identification: A Benchmark2015
- Saliency map based improved segmentation2015
- Going deeper with convolutions2015
- Saliency Detection: A Boolean Map Approach2013
- ImageNet: A large-scale hierarchical image database2009