"""
Trainer to learn depth information on unlabeled data (raw images/videos)

Allows pluggable depth networks for differing performance (including fast-depth)
"""