Abstract
Recent advances in 3D reconstruction enable high-quality rendering of complex scenes, yet they struggle to recover detailed meshes for specific targets. OMEGAS introduces a Gaussian-splatting-based segmentation pipeline that extracts 3D-consistent target masks and bootstraps mesh reconstructions before replenishing unseen regions via diffusion priors. Experiments show that the method recovers target geometry with greater precision than existing large-scene reconstruction pipelines.
Publication
arXiv preprint arXiv:2404.15891
Pu Cao
Ph.D. student of Artificial Intelligence
I’m a second-year Ph.D. student studying at Beijing University of Posts and Telecommunications (BUPT) under the supervision of Prof. Qing Song and Dr. Lu Yang. I am now interested in Computer Vision and am currently working on Image Generaion.