- Updated Python packaging to include configspec.ini and config.ini as data files in the package and updated code to use installed .ini files
- Fixed bug in format of submission.csv (was missing header row)
- Created a PyTorch Lightning data module to work with the Kaggle Global Wheat Detection dataset
- Set up a PyTorch Lightning model to wrap the torchvision Faster R-CNN object detector
- Set up to allow creating a pip installable Python package
- Created Jupyter notebooks for dataset visualization, training, and evaluation
- Created Python entry points for training, evaluation, and inference
- Added cyclic learning rate scheduler