Let's Talk Research: Going from 2D to 3D (CVLab)

​LauzHack presents its new initiative: "Let's Talk Research".

Have you ever wondered how research is done? Do you want to do research but think it is hard and too complicated? Or just want to get into a new research topic?

Join us for the "Let's Talk Research" events and we will show you that you can do research too!

---

Description: 2D data (i.e. images) is fairly easy to represent and process, but what about 3D? In this talk, we will present the intrinsic challenges of 3D data, and how to overcome them. How do you go from an image or a video to a whole 3D reconstruction of an object? How do you even represent a 3D object? How do neural models reason about 3D objects in space? In this talk, we will describe the challenges posed by 3D data and we will present the most fundamental developments in the research about how to tackle these challenges, starting from basic topics such as simple multi-layer perceptrons.

Federico, Saqib, and Chen will present the research they do in their lab, including required topic-specific background. They will also tell you about the projects you may be interested in doing in CVLab.

---

Requirements: basic understanding of deep learning (Back-propagation; linear/convolutional layers; RNNs/Transformers). If you are not familiar with these techniques or want to refresh your memory, check our Deep Learning bootcamp GitHub and Recordings (Week01 is enough).