Computer Vision News - July 2018

Adriana, you are organizing a workshop tomorrow. This is a workshop on understanding visual advertisements, which is a very new and challenging task. There’s not that much work in understanding advertisements so far. The point of this workshop is to bring together people who are, not necessarily working on this problem, but working on methods which can help in understanding ads. What I mean by understanding advertisements is to understand what the ad is trying to convince you to do, whether that’s to buy a car or help protect the environment, and what arguments it provides visually to convey that message, which can be very challenging from a vision perspective, but also a reasoning perspective. Is it an idea of yours? Working on advertisements came from my background in media studies. I did a double Computer Science and Media Studies major in college - Pomona College in California. I’ve always been interested in art and how the media makes us think, how it plays on our beliefs and maybe biases. The workshop was kind of a bold idea, since - being the only organizer - I was afraid that it would be a lot of work. It ended up being fine. I had help from a great undergraduate student as well. What take-home thoughts do you expect from tomorrow’s workshop? I don’t know yet! [ smiles ] That’s partly why I’m so looking forward to the workshop. We also have these brainstorming sessions, which I’m going to kind of moderate. Some of them are going to be game-like. Hopefully, we’ll have good attendance so we can have great brainstorming sessions. Did you invite anyone in advertising? That’s a great question! I haven’t reached out to advertisers specifically, partly because I haven’t thought of a good way to do it. I have been talking to someone in political science in my department. I think that would be greatly helpful. The reason I haven’t done it so much is because I feel like the challenges are more in the realm of AI. I haven’t had strong enough reasons to reach out to advertisers. I take personal offense because I could have been one! [ both laugh ] Maybe we can collaborate in the future. What do you believe, ideally, will be the resulting effects of the workshop tomorrow? There are several. On the one hand, I believe that understanding advertisements and persuasion in the media is important because people are exposed to images in the media which have implicit or explicit persuasive aspects. Adriana Kovashka Adriana Kovashka is an Assistant Professor in Computer Science at the University of Pittsburgh. 34 Women in Computer Vision Thursday