Skip to content

Unsupervised Data Augmentation implemented in PyTorch

Notifications You must be signed in to change notification settings

AlexTheKing/AnimalsUDA

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

1 Commit
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Preambula

This repository contains code to train the image classification model using the semi-supervised approach known as Unsupervised Data Augmentation

I ported the initial implementation for images in Tensorflow to PyTorch, so this trainer uses:

  • RandAugment to augment image data (using albumentations to improve the performance)
  • Confidence-based masking
  • Training Signal Annealing
  • Predictions Sharpening using Softmax temperature

The code was primarily developed for the Wildlife Image Classification Competition, but may be pretty easily adapted to any other image classification task.

About

Unsupervised Data Augmentation implemented in PyTorch

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages