INTERESTING.JPG explores the capabilities and limitations of the current state-of-the-art in image analysis and feature detection. Working with a model developed by the University of Toronto’s Deep Learning Group, it attempts to generate captions for images of breaking news events. It automatically posts the images and their captions to a twitter account, and source is available on github.

a bunch of men trying to climb a tree .

a view of people sitting in front of a large stone building .

the two men are talking to each other on the sidewalk .

a man riding a bike on a dirt road .

a man is taking a break from the ground .

this is a man riding a body of water .

a car is parked in front of a parking lot .

a group of young girls are looking at the airport .

a couple of people dressed in costume .

a group of men in suits and ties .

two workers are being lifted out of a pile of metal machinery .

a large group of people that are sitting next to each other .

a gymnast is performing a stunt in a red shirt and white .

a man sitting down with a smile in his hand .

a white dog wearing a collar holding a tennis ball in its mouth .

a child sitting next to a tree in a field .

a large body of water in front of a lake .

a man wearing a black shirt is looking at something art .

a group of men sitting on what appears to be a field .

three people standing on top of a large crowd .