GridMask Data Augmentation

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Chen, Pengguang, Liu, Shu, Zhao, Hengshuang, Wang, Xingquan, Jia, Jiaya
Format: Preprint
Published: 2020
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866910312804909056
author Chen, Pengguang
Liu, Shu
Zhao, Hengshuang
Wang, Xingquan
Jia, Jiaya
author_facet Chen, Pengguang
Liu, Shu
Zhao, Hengshuang
Wang, Xingquan
Jia, Jiaya
contents We propose a novel data augmentation method `GridMask' in this paper. It utilizes information removal to achieve state-of-the-art results in a variety of computer vision tasks. We analyze the requirement of information dropping. Then we show limitation of existing information dropping algorithms and propose our structured method, which is simple and yet very effective. It is based on the deletion of regions of the input image. Our extensive experiments show that our method outperforms the latest AutoAugment, which is way more computationally expensive due to the use of reinforcement learning to find the best policies. On the ImageNet dataset for recognition, COCO2017 object detection, and on Cityscapes dataset for semantic segmentation, our method all notably improves performance over baselines. The extensive experiments manifest the effectiveness and generality of the new method.
format Preprint
id arxiv_https___arxiv_org_abs_2001_04086
institution arXiv
publishDate 2020
record_format arxiv
spellingShingle GridMask Data Augmentation
Chen, Pengguang
Liu, Shu
Zhao, Hengshuang
Wang, Xingquan
Jia, Jiaya
Computer Vision and Pattern Recognition
We propose a novel data augmentation method `GridMask' in this paper. It utilizes information removal to achieve state-of-the-art results in a variety of computer vision tasks. We analyze the requirement of information dropping. Then we show limitation of existing information dropping algorithms and propose our structured method, which is simple and yet very effective. It is based on the deletion of regions of the input image. Our extensive experiments show that our method outperforms the latest AutoAugment, which is way more computationally expensive due to the use of reinforcement learning to find the best policies. On the ImageNet dataset for recognition, COCO2017 object detection, and on Cityscapes dataset for semantic segmentation, our method all notably improves performance over baselines. The extensive experiments manifest the effectiveness and generality of the new method.
title GridMask Data Augmentation
topic Computer Vision and Pattern Recognition
url https://arxiv.org/abs/2001.04086