Google’s Show and Tell algorithim is capable of producing automatic descriptive captions of images with incredible accuracy. Google has made the “latest version of our image captioning system available as an open source model in TensorFlow” if you want to try it. They’re reporting a 94% accuracy in the system accurately describing what is taking place in the images. They’ve been researching and perfecting the system for years and now if you want to play around with the source code, you’re more than welcome to HERE.