Many people create and share pictures and videos on the Internet. The difficulty is finding what you want. Typically, people want to search using words (rather than, say, example sketches). Because most pictures don’t come with words attached, it is natural to try and build tagging systems that tag images with relevant words. The underlying machinery is straightforward ― we apply image classification and object detection methods and tag the image with the output words. But tags aren’t __________. It matters who is doing what, and tags don’t capture this. For example, tagging a picture of a cat in the street with the object categories “cat”, “street”, “trash can” and “fish bones” leaves out the information that the cat is pulling the fish bones out of an open trash can on the street.