Ataman, Fikriye2025-09-302025-09-3020259798369398180979836939816610.4018/979-8-3693-9816-6.ch0102-s2.0-105013699107https://doi.org/10.4018/979-8-3693-9816-6.ch010https://hdl.handle.net/20.500.14720/28575Image data augmentation is a technique for artificially expanding and diversifying existing image datasets. This method is beneficial when training data is insufficient for machine learning and deep learning models. It allows models to generalize better during training and reduces overfitting. The main techniques include geometric transformations, color and light changes, noise addition, optical distortions, and modern blending methods. These methods expose the model to more diversity by simulating real-world variations. Thus, data collection and labeling costs are reduced while the model's performance and generalization capacity increase. Data augmentation plays a role in the basis of success, especially in studies where medical images are processed. This study explains the techniques used for image augmentation in detail. It also reveals the performance results of the methods used on data sets and their effects on deep models. The study includes the most frequently used methods and provides a detailed literature review. © 2025 Elsevier B.V., All rights reserved.eninfo:eu-repo/semantics/closedAccessDeep LearningLearning SystemsMathematical TransformationsMedical Image ProcessingPersonnel TrainingAugmentation TechniquesData AugmentationGeometric TransformationsImage DataImage DatasetsImage ProcessingLearning ModelsMachine LearningOverfittingTraining DataBlendingBook PartN/AN/A231262