0 / 60 seg.

The more training data it's given, the better the algorithm becomes at correctly identifying new images.