EE Seminar: On the Role of Geometry in Geo-Localization
Speaker: Moti Kadosh
M.Sc. student under the supervision of Prof. Ariel Shamir and Prof. Daniel Cohen Or
Wednesday, February 27th, 2019 at 15:00
Room 011, Kitot Bldg., Faculty of Engineering
On the Role of Geometry in Geo-Localization
Abstract
Humans can build a mental map of a geographical area to find their way and recognize places. The basic task we consider is finding the pose (position & orientation) of a camera in a large 3D scene from a single image. Our goal is to examine whether such a capability can be learned by a neural network. In particular, we aim to explore the role of geometry alone in geo-localization using CNNs, while ignoring the often available texture of the scene. We therefore deliberately avoid using texture or rich geometric details and use images projected from a simple 3D model of a city, which we term lean images. Lean images contain mostly information that relates to the geometry of the area viewed (edges, faces, or relative depth). We find that the network is capable of estimating the camera pose from the lean images, not by memorization but by some measure of geometric learning of the geographical area. The main contributions of this thesis are: (i) demonstrating the power of CNNs for recovering camera pose using lean images; and (ii) providing insight into the role of geometry in the CNN learning process.
