A survey on Image Data Augmentation for Deep Learning
By Connor Shorten, Taghi M. Khoshgoftaar. 2019
11,876 citations recorded by the source corpus.
Deep convolutional neural networks have performed remarkably well on many Computer Vision tasks. However, these networks are heavily reliant on big data to avoid overfitting. Overfitting refers to the phenomenon when a network learns a function with very high variance such as to perfectly model the training data. Unfortunately, many application domains do not have access to big data, such as medical image analysis. This survey focuses on Data Augmentation, a data-space solution to the problem of limited data. Data Augmentation encompasses a suite of techniques that enhance the size and quality of training datasets such that better Deep Learning models can be built using them. The image augmentation algorithms discussed in this survey include geometric transformations, color space augmentations, kernel filters, mixing images, random erasing, feature space augmentation, adversarial training, generative adversarial networks, neural style transfer, and meta-learning. The application of augmentation methods based on GANs are heavily covered in this survey. In addition to augmentation techniques, this paper will briefly discuss other characteristics of Data Augmentation such as test-time augmentation, resolution impact, final dataset size, and curriculum learning. This survey will present existing methods for Data Augmentation, promising developments, and meta-level decisions for implementing Data Augmentation. Readers will understand how Data Augmentation can improve the performance of their models and expand limited datasets to take advantage of the capabilities of big data.
Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Representation & Self-Supervised Learning
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Quoc V. Le
- Mohit Bansal
- Jianfeng Gao
- Yoshua Bengio
- Percy Liang
- Sergey Levine
- Caiming Xiong
- Trevor Darrell
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Xiaogang Wang
- Trevor Darrell
- Alan L. Yuille
- Jitendra Malik
- Andrew Zisserman
In Representation & Self-Supervised Learning
- Quoc V. Le
- Sergey Levine
- Caiming Xiong
- Steven C. H. Hoi
- Ekin Dogus Cubuk
- Chelsea Finn
- Junnan Li
- Barret Zoph
Citations in the corpus
- Training Generative Adversarial Networks with Limited Data (2020, background): “In almost all areas of deep learning [40], dataset augmentation is the standard solution against overfitting.”
- Generative Adversarial Networks (2021, neutral): “This approach has been successfully applied in various areas s.a. object detection, image classification, and speech recognition [18].”
- A Survey of Data Augmentation Approaches for NLP (2021, background): “types of data, such as images (Shorten and Khoshgoftaar, 2019), faces (Wang et al.”
- BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation (2022, background): “While data augmentation (DA) has been widely adopted in computer vision (Shorten & Khoshgoftaar, 2019), DA for language tasks is less straightforward.”
- Simple Copy-Paste is a Strong Data Augmentation Method for Instance Segmentation (2020, methodology): “Compared to the volume of work on backbone architectures [35, 51, 53, 27, 56] and detection/segmentation frameworks [19, 18, 47, 38, 26, 39], relatively less attention is paid to data augmentations [50] in the computer vision community.”
- ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision (2021, background): “Image augmentation reportedly improves the generalization power of vision models (Shorten & Khoshgoftaar, 2019).”
- Pyramid Vision Transformer: A Versatile Backbone for Dense Prediction without Convolutions (2021, background): “More recent advances can be found in recent survey/review papers [30, 52].”
- VEGA: Towards an End-to-End Configurable AutoML Pipeline (2020, background): “There are many surveys (He et al., 2019; Z¨oller & Huber, 2019; Elsken et al., 2019; Cheng et al., 2017; Shorten & Khoshgoftaar, 2019) that have summarized classical algorithms on these topics.”
- Widening the Pipeline in Human-Guided Reinforcement Learning with Explanation and Context-Aware Data Augmentation (2020, methodology): “Data augmentation techniques are widely used in computer vision tasks [21, 40] and have recently been applied to deep RL tasks [18, 20, 33, 26].”
- A Survey on Data Augmentation for Text Classification (2021, neutral): “While other methods reduce the bias error, data augmentation’s objective is to keep it constant and is used to solve the problem at the root [5].”
- Medical image data augmentation: techniques, comparisons and interpretations (2023, background): “Therefore, translated images provide prevention of positional bias (Shorten and Khoshgoftaar 2019) and the models do not focus on properties in a single spatial location (Nalepa et al. 2019).”
- Deep semantic segmentation of natural and medical images: a review (2019, background): “However, the majority of GAN based data augmentation has been applied to medical images (Shorten and Khoshgoftaar, 2019).”
- A review of medical image data augmentation techniques for deep learning applications (2021, background): “However, these articles mainly focus on generative adversarial networks (GANs) based augmentation,(9,10) a specific task of either image classification or segmentation,(6,7) or general data augmentation for computer vision.(8) To the best of our knowledge, this work is the first to provide a compreh”
- Characterizing & Finding Good Data Orderings for Fast Convergence of Sequential Gradient Methods (2022, background): “For example, consider an image classification task where random rotations of data points are added to the dataset for augmentation purposes which is a widely used technique [27].”
- Deblurring by Realistic Blurring (2020, methodology): “The idea of data augmentation has been widely applied in different fields [31], like face verification [20] and SR [1].”