Computer Vision News - March 2023

4 AI Research also been utilized in robotics for bundle adjustment, city planning, and coordination between multiple robots. In 2D, you can easily perform a basic operation of rigid registration between two images , making the same objects in both images fit together by aligning the two with some rotation and translation. This work aims to achieve the same functionality on NeRFs. Where there are two 3D scenes with different lighting and backgrounds but With the growing popularity of neural radiance fields (NeRFs) as 3D scene representations, there are an increasing number of datasets and databases of NeRFs from different scenes which need to be processed, as is already the case for 2D images. Registration is a technique widely used in image processing, particularly in medical applications for detecting longitudinal changes in images, such as MRIs. It has NERF2NERF: PAIRWISE REGISTRATION OF NEURAL RADIANCE FIELDS Lily Goli is a second-year PhD student at the University of Toronto, co-supervised by Andrea Tagliasacchi and Alec Jacobson. Her paper proposing a new technique for registering neural radiance fields has been accepted at ICRA2023, anupcoming robotics conference in London. Lily is here to tell us what it is all about.