Computer Vision News - January 2024

Computer Vision News 2 NeurIPS 2023 Paper Denys Rozumnyi is a fifth-year PhD student at ETH Zurich under the supervision of Marc Pollefeys. In his research, he works on object-centric 3D reconstruction, motion estimation, deblurring, and detection. In particular, he focused a lot on motion-blurred objects. by Denys Rozumnyi Estimating 3D structures from video is one of the key tasks in computer vision. The key ingredient of state-of-the-art methods are large annotated datasets for training. Some datasets even offer 3D room layout, which is defined as a set of 3D structural elements such as wall, floor, and ceiling, but there are just a few real datasets with annotated 3D room layouts. However, they require the images/videos to be acquired with special devices or sensors such as RGB-D cameras or panoramic captures. Estimating Generic 3D Room Structures from 2D Annotations Matias Valdenegro