{"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/mid-fusion-octree-based-object-level-multi","title":"MID-Fusion: Octree-based Object-Level Multi-Instance Dynamic SLAM","arxiv_id":"1812.07976","date":"2018-12-19","proceeding":null,"authors":["Binbin Xu","Wenbin Li","Dimos Tzoumanikas","Michael Bloesch","Andrew Davison","Stefan Leutenegger"],"abstract":"We propose a new multi-instance dynamic RGB-D SLAM system using an\nobject-level octree-based volumetric representation. It can provide robust\ncamera tracking in dynamic environments and at the same time, continuously\nestimate geometric, semantic, and motion properties for arbitrary objects in\nthe scene. For each incoming frame, we perform instance segmentation to detect\nobjects and refine mask boundaries using geometric and motion information.\nMeanwhile, we estimate the pose of each existing moving object using an\nobject-oriented tracking method and robustly track the camera pose against the\nstatic scene. Based on the estimated camera pose and object poses, we associate\nsegmented masks with existing models and incrementally fuse corresponding\ncolour, depth, semantic, and foreground object probabilities into each object\nmodel. In contrast to existing approaches, our system is the first system to\ngenerate an object-level dynamic volumetric map from a single RGB-D camera,\nwhich can be used directly for robotic tasks. Our method can run at 2-3 Hz on a\nCPU, excluding the instance segmentation part. We demonstrate its effectiveness\nby quantitatively and qualitatively testing it on both synthetic and real-world\nsequences.","url_abs":"http://arxiv.org/abs/1812.07976v4","url_pdf":"http://arxiv.org/pdf/1812.07976v4.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":"mid-fusion-octree-based-object-level-multi","repo_url":"https://github.com/smartroboticslab/mid-fusion","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"BSD-3-Clause"}}],"tasks":[{"task_slug":null,"task_name":"CPU"},{"task_slug":"instance-segmentation","task_name":"Instance Segmentation"},{"task_slug":"object","task_name":"Object"},{"task_slug":"object-slam","task_name":"Object SLAM"},{"task_slug":"semantic-slam","task_name":"Semantic SLAM"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1812.07976","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}