1. Generative Modeling Method Comparison for Class Imbalance Correction

An image dataset, for use in object detection of hay bales, with over 6000 images of both good and bad hay bales was collected.  Unfortunately, the dataset developed a class imbalance, with more good bale images than bad bales.  This dataset class imbalance caused the bad bale class to over train and the good bale class to under train, severely impacting precision, and recall.  To correct this imbalance and provide a comparison of differing generative modeling methods; three different... B. Vail, Z. Oster, B. Weinhold