Computer Vision News - December 2019

Women in C. Vision 28 Georgia Gkioxari is a research scientist at Facebook AI Research (FAIR). She received a PhD in computer science and electrical engineering from the University of California at Berkeley under the supervision of Jitendra Malik in 2016. Her research lies in computer vision with a focus on object and person recognition from static images and videos. Georgia received the Marr Prize at ICCV for "Mask R-CNN" in 2017. Georgia, tell us about your work. I can start with the 3D stuff that we’ve been working on. We actually had a tutorial on Monday. We presented our work on Mesh R-CNN, which is basically inspired by Mask R-CNN, but trying to move beyond the 2D grid, a grid we are very comfortable in. We want to predict in the 3D, get the shapes of objects, not just the silhouettes of objects. Then we also presented torch3d, a library of 3D operations used in Mesh R-CNN and beyond, including a differentiable renderer. It’s particularly designed to enable and facilitate research in deep learning and 3D understanding. This will all be published very soon. We hope that the community will appreciate the work and build on top of these tools. You work in a very celebrated team with very special people. That must be challenging and rewarding at the same time. Definitely - My colleagues have set a very, very high bar, not just for me, but for everyone at this conference. I’m very proud to work with them. When I work on a project, I have to think, “What will my colleagues think?” If I know that they will be proud of this work, then I know it’s good work. I care more about their approval than I care about CVPR or ICCV being in my favor. How do you deal with the pressure? I don’t think I feel pressure every day because, if you meet my team, they’re super normal. I don’t feel a particular pressure every day, but I think, overall, when I’m finalizing a project or when I’m writing up a project, I think, “How If there is no trust, nothing will come“ "I love dogs! They don't always love me back!" Best of ICCV 2019

RkJQdWJsaXNoZXIy NTc3NzU=