CVPR Daily - Monday

Post-processing combines these local patches so that they are consistent with each other. “ In the past, you had many representations of shapes like point clouds and meshes, ” Marie-Julie tells us. “ What’s really exciting is that now we have this great tool in deep learning. We can combine these two exciting fields to get even more interesting methods that can adapt to any kind of data for which we will have these amazing applications! ” There are many applications for this work, including enhanced 3D modelling and mapping . Looking to the future, Marie-Julie says the ultimate goal is to have a fully differentiable end-to-end pipeline that can input a representation of a 3D shape and reconstruct the point cloud and the mesh at the same time without any post-processing. Finally, we can’t let Marie-Julie go without letting her know one more reason why we are so thrilled to interview her. We have spoken to many African scientists for our magazine, but she happens to be our very first Malagasy scholar! We ask if she can tell us something about her home country Madagascar that we would not know. She ponders the question for a moment. “ I don’t want to say something cliché like there are beautiful beaches or things like that! ” she laughs. “ Okay, so we have this dish which is called ravitoto . It’s very green and it looks disgusting. Honestly, Google it. It looks terrible. But I promise you, it’s delicious! ” To learn more about Marie- Julie’s work, you can view her presentation and PDF poster anytime, and take part in her live Q&A during Paper Session One today (Monday) at 11:00-13:30 EDT. 15 DAILY CVPR Monday Marie-Julie Rakotosaona