{"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/learning-3d-shape-completion-under-weak","title":"Learning 3D Shape Completion under Weak Supervision","arxiv_id":"1805.07290","date":"2018-05-18","proceeding":null,"authors":["David Stutz","Andreas Geiger"],"abstract":"We address the problem of 3D shape completion from sparse and noisy point\nclouds, a fundamental problem in computer vision and robotics. Recent\napproaches are either data-driven or learning-based: Data-driven approaches\nrely on a shape model whose parameters are optimized to fit the observations;\nLearning-based approaches, in contrast, avoid the expensive optimization step\nby learning to directly predict complete shapes from incomplete observations in\na fully-supervised setting. However, full supervision is often not available in\npractice. In this work, we propose a weakly-supervised learning-based approach\nto 3D shape completion which neither requires slow optimization nor direct\nsupervision. While we also learn a shape prior on synthetic data, we amortize,\ni.e., learn, maximum likelihood fitting using deep neural networks resulting in\nefficient shape completion without sacrificing accuracy. On synthetic\nbenchmarks based on ShapeNet and ModelNet as well as on real robotics data from\nKITTI and Kinect, we demonstrate that the proposed amortized maximum likelihood\napproach is able to compete with recent fully supervised baselines and\noutperforms data-driven approaches, while requiring less supervision and being\nsignificantly faster.","url_abs":"http://arxiv.org/abs/1805.07290v2","url_pdf":"http://arxiv.org/pdf/1805.07290v2.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":"learning-3d-shape-completion-under-weak","repo_url":"https://github.com/davidstutz/ijcv2018-improved-shape-completion","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"learning-3d-shape-completion-under-weak","repo_url":"https://github.com/davidstutz/aml-improved-shape-completion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"learning-3d-shape-completion-under-weak","repo_url":"https://github.com/davidstutz/cvpr2018-shape-completion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"learning-3d-shape-completion-under-weak","repo_url":"https://github.com/davidstutz/mesh-fusion","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"learning-3d-shape-completion-under-weak","repo_url":"https://github.com/paschalidoud/mesh_fusion_simple","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}}],"tasks":[{"task_slug":"weakly-supervised-learning","task_name":"Weakly-supervised Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1805.07290","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}