Előadás címe: The connection of 3D geometry and first-order correspondence
Helyszín: Online, https://meet.google.com/ete-gaku-aex
Időpont: 2020.06.29., 16:50 – 17:10
Kivonat: In computer vision, mainstream approaches only consider point correspondences to solve geometric problems. However, corresponding features usually contain more useful information about the underlying geometry. Recent works aim to exploit this additional information to facilitate the solution of geometric computer vision problems. Camera-based 3D reconstruction and SLAM systems heavily rely on robust geometric estimation to determine camera motion and a representation of the environment. Using first order correspondences, such an approach can benefit from reduced computational demand and a better exploration of the parameter space. In this presentation, first the relevant theoretical and practical approaches are explored, followed by our related project goals and our latest advancements.