Did several experiments on using Google Cloud AI integration.
One of them is selecting random keyframes from the VRT Production Asset Management System by using some PHP scripting and then uploading and putting these images through the Google Cognitive Cloud AI services. By using facial recognition, you get some coordinates of where Google recognizes a person in each image. I used these coordinates to put a green square around the persons faces on each of the keyframes and displayed these modified images in an animated three.js image cloud, just because it looks way cooler that way ;-). Check it out at