Skip to content

Latest commit

 

History

History
36 lines (28 loc) · 1.63 KB

File metadata and controls

36 lines (28 loc) · 1.63 KB

Data Augmentation for Deep Learning

Automatic Data Augmentation for Deep Learning techniques.

Goal

You can generate new data to train neural networks. This is an easy way to prevent the model overfitting.

Features

  • Crop
  • Rotate
  • Noise (Gaussian Noise added)
  • Flip-up
  • Change brightness

Examples

Original Crop Rotate
Noise Flip-up Change brightness

Requirements

Software Version Required
Python >= 3.5 Yes
Numpy >= 1.13 Yes
opencv-python >= 3.4.2.17 Yes
os - Yes
logging - Yes
imageio >= 2.9.0 Yes

Credits

Thanks to Alex Turner