A render visualising how humans can teach machine learning models to identify false positives and curate training data.