A Comprehensive Survey of Image Augmentation Techniques for Deep Learning

• Published in: Pattern recognition Vol. 137; p. 109347
• Main Authors: Xu, Mingle, Yoon, Sook, Fuentes, Alvaro, Park, Dong Sun
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.05.2023
• Subjects: Computer vision; Data augmentation; Deep learning; Image augmentation; Image variation; Vicinity distribution; Deep learning; Computer vision; Data augmentation; Image augmentation; Image variation; Vicinity distribution
• ISSN: 0031-3203; 1873-5142
• DOI: 10.1016/j.patcog.2023.109347

•We examine challenges and vicinity distribution to demonstrate the necessity of image augmentation for deep learning.•We present a comprehensive survey on image augmentation with a novel informative taxonomy that encompasses a wider range of algorithms.•We discuss the current situation and future direction for image augmentation, along with three relevant topics: understanding image augmentation, new strategy to leverage image augmentation, and feature augmentation. Although deep learning has achieved satisfactory performance in computer vision, a large volume of images is required. However, collecting images is often expensive and challenging. Many image augmentation algorithms have been proposed to alleviate this issue. Understanding existing algorithms is, therefore, essential for finding suitable and developing novel methods for a given task. In this study, we perform a comprehensive survey of image augmentation for deep learning using a novel informative taxonomy. To examine the basic objective of image augmentation, we introduce challenges in computer vision tasks and vicinity distribution. The algorithms are then classified among three categories: model-free, model-based, and optimizing policy-based. The model-free category employs the methods from image processing, whereas the model-based approach leverages image generation models to synthesize images. In contrast, the optimizing policy-based approach aims to find an optimal combination of operations. Based on this analysis, we believe that our survey enhances the understanding necessary for choosing suitable methods and designing novel algorithms.