TODO.md

Todo List

• Distribution

  • scrap images recursively from a given path
  • generate stats about total images under each sub folder
  • calculate total augmenations to make in order to balance data
  • plot a BAR + PIE chart
  • encapsulate everything under one function ft_distribution(target_path, totalVariants, plot_chart=False)
  • make the function accesible via terminal run e.g: python3 000-Distribution.py ./dataset/Apple
  • split data between validation/train datasets
    • specify flag for it

• Augmenation

  • make a an ImageAugmentor class to generate all possible 7 augmentations
  • encapsulate all in single function
  • save the augmented images into same folder as original
  • use ft_distribution to augment all images inside given path
  • save augmented images in augmented_directory by default or from user by args
  • plot augmented images if a single image is given as argument

• Transformation

  • make the transformation functions:
    • guassian_blur
    • mask
    • roi (range of interests) objects
    • analysis object
    • pseudo-landmarks
    • colors historgram
  • if given path is image transform it and show output in place
  • if given path is directory then transform contained images recursively with their class name to destination project