Results look a bit uninspiring, but you can see the potential for bigger/more refined datasets and more training.