simulated+unsupervised (s+u) learning

An approach that improves the realism of synthetic images from a simulator using unlabeled real data, helping to bridge the 'reality gap'.