Given a series of images in multiple domains, such as RGB, depth and normal, create augmented datasets by applying transformations, such as rotation and blur, to images in each domain. For more details, see this link.
conda create -n AUG python=3.11
conda activate AUG
python setup.py develop
Before running the script, download the sample data from this link and save under data or a path of your choice.
python augmentor.py --config-file path/to/config
Config parameters are defined as a YAML file for augmentor
Input images are loaded as data_augmentation.loader.Dataset
Image transforms currently supported are defined in data_augmentation.transforms
To apply a set of image Transform to input Dataset, use the data_augmentation.augment.Augment
Go to sample_annotations directory for sample annotations generated using augmentor.py