Data augmentation

AI training technique that generates image variants to enrich the dataset.

📖 Full definition

To train a robust model, you artificially generate variants of the original images: rotations, zooms, blur, brightness shifts, mirroring. This lets the AI learn to recognise a face despite these natural variations, without needing millions of extra examples.

💡 Concrete example

A single photo of you, augmented 20 times (rotation, light, mirror), produces 20 training examples for the model.

🔗 See also

CNN (Convolutional Neural Network) →
Deep learning →
Transfer learning →

🪞 Test resemblance with our AI

Start my free test →