0 / 60 seg.

So what we thought is, let's strip out all the metadata, let's look at what machine learning can do based purely on visual recognition of this entire collection.