Skip to Content

This is a demo from the University of Toronto's deep learning group. The results of a machine vision system producing descriptions of images are posted to Twitter. The results are often humorous, but an excellent illustration of the difficulty of contextual interpretation, even when objects are correctly identified.