Active learning (AL) is an increasingly popular strategy for mitigating the amount of labeled data required to train classifiers, thereby reducing annotator effort. We describe a real-world, deplo