[Paper Review] Image Data Augmentation for Deep Learning: A Survey

Suorong Yang, Weikang Xiao+4 more|arXiv (Cornell University)|Apr 19, 2022

Advanced Neural Network ApplicationsComputer Science198 citations

TL;DR

A comprehensive survey of image data augmentation methods, taxonomy, and empirical evaluation across semantic segmentation, image classification, and object detection.

ABSTRACT

Deep learning has achieved remarkable results in many computer vision tasks. Deep neural networks typically rely on large amounts of training data to avoid overfitting. However, labeled data for real-world applications may be limited. By improving the quantity and diversity of training data, data augmentation has become an inevitable part of deep learning model training with image data. As an effective way to improve the sufficiency and diversity of training data, data augmentation has become a necessary part of successful application of deep learning models on image data. In this paper, we systematically review different image data augmentation methods. We propose a taxonomy of reviewed methods and present the strengths and limitations of these methods. We also conduct extensive experiments with various data augmentation methods on three typical computer vision tasks, including semantic segmentation, image classification and object detection. Finally, we discuss current challenges faced by data augmentation and future research directions to put forward some useful research guidance.

Motivation & Objective

• Motivate the study by addressing data scarcity and the need for diverse training data in deep learning for vision.
• Propose a taxonomy of image data augmentation methods.
• Systematically review augmentation techniques across key CV tasks (segmentation, classification, detection) and compare their effectiveness.
• Provide empirical evaluations on standard datasets to assess augmentation methods and guidance for future research.

Proposed method

• Propose a taxonomy of image augmentation methods, including basic manipulations, erasing, and mixing.
• Group advanced techniques into auto-augmentation, feature augmentation, and deep generative models.
• Conduct extensive experiments on semantic segmentation, image classification, and object detection using public datasets (e.g., PASCAL VOC, COCO) to compare augmentation methods.
• Discuss theoretical and evaluation challenges, and propose future research directions.

Experimental results

Research questions

• RQ1 What augmentation techniques exist for image data across CV tasks and how are they categorized?
• RQ2 How do different augmentation methods affect performance in segmentation, classification, and detection tasks on standard datasets?
• RQ3 What are the limitations and challenges (theory, evaluation, data quality) in current augmentation approaches?
• RQ4 What future directions can improve the effectiveness and understanding of data augmentation?

Key findings

• Data augmentation consistently improves performance across models and tasks (e.g., IoU gains in segmentation, accuracy gains in classification, mAP improvements in detection).
• A broad taxonomy captures basic manipulations, erasing, and mixing, plus advanced approaches like auto augmentation, feature augmentation, and GAN-based methods.
• AutoAugment, RandAugment, and related strategies reduce search cost and improve effectiveness compared to manual design.
• Evaluation on standard datasets shows notable gains across CNN architectures and vision tasks.
• Open challenges include theoretical understanding, evaluation metrics for synthetic data quality, and dataset size considerations.
• Combinations of augmentation methods often outperform single-method setups.