We need a way to store and load trained segmentation models. Currently, we are using scikit-learn to train models, which suggests a couple of ways to serialize the model via pickle and joblib. https://scikit-learn.org/stable/model_persistence.html