This project contains two scripts for image augmentation using the Albumentations library and okankop/vidaug. This is meant to be used with the following GitHub Repository: https://github.com/hjl20/issp3900-handwash-annotation

This script currently loads an image and its corresponding YOLO bounding box annotations, applies a sequence of augmentation techniques to the image and updates the bounding box annotations accordingly, and saves the augmented image and updated annotations to a new directory.

To use this script, you need to have Python installed on your machine along with the cv2 and albumentations libraries.

1. Replace the label_file and image_file variables with the paths to your YOLO bounding box annotations and image respectively.
2. Uncomment or add the augmentations you want to apply inside the A.Compose([]) function.
3. Run the script. It will load the image and its annotations, apply the augmentations, update the annotations, and save the augmented image and updated annotations to the specified directory.

Refer to this website for additional augmentations: https://albumentations.ai/docs/api_reference/augmentations/ The script available augmentations from the albumentations library. Note that some augmentations, like flips, crops, or resizes, may require changes to any bounding boxes associated with the image.

This script loads a batch of images from a specified directory, applies a sequence of augmentation techniques to each image, and saves the augmented images to a new directory.

To use this script, you need to have Python installed on your machine along with the cv2, vidaug, and skimage libraries.

1. Replace the image_folder variable with the path to your images.
2. Replace the augmented_dir variable with the path where you want to save the augmented images.
3. Uncomment or add the augmentations you want to apply inside the va.Sequential([]) function.
4. Run the script. It will load each image, apply the augmentations, and save the augmented images to the specified directory.

Refer to this website for augmentation implementation:https://github.com/okankop/vidaug The script includes a list of all available augmentations from the vidaug library. Note that some augmentations, like flips, crops, or resizes, may require changes to any bounding boxes associated with the image.