This is very interesting for people like me who don’t want to learn everything about Tensorflow or Caffe but still want to see what’s the current state of affairs and have fun with the tech.
For example I spent my weekend running places/scenes recognition models and objects recognition models on my Lightroom pictures library, all on my laptop, with a tiny script adding the recognized labels to the pictures as IPTC keywords using exiftools. Quite some manual wiring work but I can now filter my pictures by content! (Results aren’t very impressive though. I’ll try other models!)