Computer Vision News - November 2020

242 It would be him dancing and not Michael Jackson dancing? Yes, in a way it's still a puppet because the motion is coming from the other person, but the appearance is all this person. The source and the target. We are doing this several frames at a time, which gives you the motion. I think the interesting thing about this one is that afterwards, a lot of people said, “You should really be able to animate somebody from a single image. Take an image and make them dance.” That’s nice, but you don't ever have any chance to know what the back of their head looks like. That is what I think is so special about this project, which is a little bit different than what people were doing afterwards. I’m very interested in your experience working with Jitendra Malik and Alyosha Efros , two teachers that myself and our readers admire very much. Can you tell us something about working with them? What makes them different? Oh, absolutely! Also, it's important for me that they are different. Alyosha is highly creative and very intuitive. He thinks of ideas and concepts in a very intuitive way. He’s very inspiring in that way. For many years, he’s talking about the cycle consistency idea. It’s a high- level thing, but, slowly it's becoming something that works very well in practice for machine learning. He is way up here, on a high level with inspiring ideas. He brings a lot of art, ideas from philosophy, ideas from literature, even music into his work. He's really a Renaissance person, as he brings all these things together. Women in Computer Vision Dissertation party! At the center of the photo is Georgia Gkioxari. At the top, Jitendra Malik and Alyosha Efros