These are companion scripts to the manuscript "Self-Supervised Maize Kernel Classification and Segmentation for Embryo Identification", which has been published in vol. 14 of Frontiers in Plant Science.
- evaluate.py takes an existing model and formatted data directory to evaluate model performance.
- pipeline.py is a vision pipeline which takes a backbone trained by self-supervision or one which is preloaded, and trains it on labeled data.
The trained models can be found at https://zenodo.org/record/7577017, along with the accompanying dataset.