计算机科学与探索 ›› 2021, Vol. 15 ›› Issue (4): 583-611.DOI: 10.3778/j.issn.1673-9418.2102015

• 综述·探索 • 上一篇    下一篇

机器视觉应用中的图像数据增广综述

林成创,单纯,赵淦森,等   

  1. 1. 华南师范大学 计算机学院,广州 510631
    2. 广东技术师范大学 电子与信息学院,广州 510665
    3. 挪威科技大学,挪威 特隆赫姆 17491
    4. 广州市云计算安全与测评技术重点实验室,广州 510631
    5. 华南师范大学唯链区块链技术与应用联合实验室,广州 510631
    6. 广东财经大学 统计与数学学院,广州 510320
  • 出版日期:2021-04-01 发布日期:2021-04-02

Review of Image Data Augmentation in Computer Vision

LIN Chengchuang, SHAN Chun, ZHAO Gansen, et al   

  1. 1. School of Computer Science, South China Normal University, Guangzhou 510631, China
    2. School of Electronics and Information, Guangdong Polytechnic Normal University, Guangzhou 510665, China
    3. Norwegian University of Science and Technology, Trondheim 17491, Norway
    4. Key Lab on Cloud Security and Assessment Technology of Guangzhou, Guangzhou 510631, China
    5. South China Normal University & VeChain Joint Lab on BlockChain Technology and Application, Guangzhou 510631, China
    6. School of Statistics and Mathematics, Guangdong University of Finance and Economics, Guangzhou 510320, China
  • Online:2021-04-01 Published:2021-04-02

摘要:

深度学习是目前机器视觉的前沿解决方案,而海量高质量的训练数据集是深度学习解决机器视觉问题的基本保障。收集和准确标注图像数据集是一个极其费时且代价昂贵的过程。随着机器视觉的广泛应用,这个问题将会越来越突出。图像增广技术是一种有效解决深度学习在少量或者低质量训练数据中进行训练的一种技术手段,该技术不断地伴随着深度学习与机器视觉的发展。系统性梳理当前图像增广技术研究,从增广对象、增广空间、标签处理和增广策略生成的角度,分析现有图像增广技术的研究范式。依据研究范式提出现有图像增广技术的分类系统,重点介绍每类图像增广研究的代表性研究成果。最后,对现有图像增广研究进行总结,指出当前图像增广研究中存在的问题及未来的发展趋势。

关键词: 深度学习, 计算机视觉, 图像增广, 数据增广, 图像增强

Abstract:

Deep learning is a promising solution for computer vision at present. To solve the computer vision problem, it requires massive and high-quality image training datasets. Collecting and accurately labeling image datasets is a very time-consuming and expensive process. As computer vision applications become more widespread, it makes this problem even more pronounced. Image augmentation technologies are technical methods to effectively solve the problem of deep learning training under the condition of small-scale or low-quality training data. These technologies are continually accompanied with the development of deep learning and computer vision. This paper first reviews these image augmentation researches from the perspective of augmentation objects, operation spaces, label processing methods, and augmentation strategies and then concludes corresponding paradigms of current image data augmentation methods. After that, this paper proposes a taxonomy for current image data augmentation guided by the above paradigms, and reviews corresponding representative methods of each image data augmentation category. Finally, this paper makes conclusions on existing image data augmentation, points out the problems existing in the current image augmentation research and presents promising directions for future research.

Key words: deep learning, computer vision, image augmentation, data augmentation, image enhancement