{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/surfelwarp-efficient-non-volumetric-single","title":"SurfelWarp: Efficient Non-Volumetric Single View Dynamic Reconstruction","arxiv_id":"1904.13073","date":"2019-04-30","proceeding":null,"authors":["Wei Gao","Russ Tedrake"],"abstract":"We contribute a dense SLAM system that takes a live stream of depth images as\ninput and reconstructs non-rigid deforming scenes in real time, without\ntemplates or prior models. In contrast to existing approaches, we do not\nmaintain any volumetric data structures, such as truncated signed distance\nfunction (TSDF) fields or deformation fields, which are performance and memory\nintensive. Our system works with a flat point (surfel) based representation of\ngeometry, which can be directly acquired from commodity depth sensors. Standard\ngraphics pipelines and general purpose GPU (GPGPU) computing are leveraged for\nall central operations: i.e., nearest neighbor maintenance, non-rigid\ndeformation field estimation and fusion of depth measurements. Our pipeline\ninherently avoids expensive volumetric operations such as marching cubes,\nvolumetric fusion and dense deformation field update, leading to significantly\nimproved performance. Furthermore, the explicit and flexible surfel based\ngeometry representation enables efficient tackling of topology changes and\ntracking failures, which makes our reconstructions consistent with updated\ndepth observations. Our system allows robots to maintain a scene description\nwith non-rigidly deformed objects that potentially enables interactions with\ndynamic working environments.","url_abs":"http://arxiv.org/abs/1904.13073v1","url_pdf":"http://arxiv.org/pdf/1904.13073v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"surfelwarp-efficient-non-volumetric-single","repo_url":"https://github.com/weigao95/surfelwarp","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"BSD-3-Clause"}}],"tasks":[{"task_slug":"dynamic-reconstruction","task_name":"Dynamic Reconstruction"},{"task_slug":null,"task_name":"GPU"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.13073","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}