# Image Augmentation

> Source: https://sukruyusufkaya.com/en/glossary/image-augmentation
> Updated: 2026-05-13T20:01:12.067Z
> Type: glossary
> Category: bilgisayarli-goru
**TLDR:** A data-driven technique that improves model generalization by diversifying training data through transformations.

<p>Image augmentation is one of the most effective regularization tools in computer vision systems trained on limited data. Many transformations can be applied, including rotation, cropping, brightness adjustment, noise injection, perspective shifts, and mix-based methods. This helps the model learn representations that are more robust to small variations in the data. It becomes especially critical in production environments where camera angle, lighting, and object position vary significantly.</p>