Home

Welcome to the data-augmentation-review wiki! Data augmentation can be simply described as any method that makes our dataset larger. To create more images for example, we could zoom the in and save a result, we could change the brightness of the image or rotate it. To get bigger sound dataset we could try raise or lower the pitch of the audio sample or slow down/speed up.

DATA AUGMENTATION

• Images

  • Affine transformations
    • Rotation
    • Scaling
    • Random cropping
    • Reflection
  • Elastic transformations
    • Contrast shift
    • Brightness shift
    • Blurring
    • Channel shuffle
  • Advanced transformations
    • Random erasing
    • Adding rain effects, sun flare...
    • Image blending
  • Neural-based transformations
    • Adversarial noise
    • Neural Style Transfer
    • Generative Adversarial Networks

• Audio

  • Noise injection
  • Time shift
  • Time stretching
  • Random cropping
  • Pitch scaling
  • Dynamic range compression
  • Simple gain
  • Equalization

• Natural Language Processing

  • Thesaurus
  • Text Generation
  • Back Translation
  • Word Embeddings
  • Contextualized Word Embeddings

• AutoAugment