It claims to have 1000 scenes of 20 seconds each....unless I'm reading that wrong, their entire dataset is less than 6 hrs of driving. A human couldn't learn i

It claims to have 1000 scenes of 20 seconds each....unless I'm reading that wrong, their entire dataset is less than 6 hrs of driving. A human couldn't learn in that time, much less a neural net.