{"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/alignet-partial-shape-agnostic-alignment-via","title":"ALIGNet: Partial-Shape Agnostic Alignment via Unsupervised Learning","arxiv_id":"1804.08497","date":"2018-04-23","proceeding":null,"authors":["Rana Hanocka","Noa Fish","Zhenhua Wang","Raja Giryes","Shachar Fleishman","Daniel Cohen-Or"],"abstract":"The process of aligning a pair of shapes is a fundamental operation in\ncomputer graphics. Traditional approaches rely heavily on matching\ncorresponding points or features to guide the alignment, a paradigm that\nfalters when significant shape portions are missing. These techniques generally\ndo not incorporate prior knowledge about expected shape characteristics, which\ncan help compensate for any misleading cues left by inaccuracies exhibited in\nthe input shapes. We present an approach based on a deep neural network,\nleveraging shape datasets to learn a shape-aware prior for source-to-target\nalignment that is robust to shape incompleteness. In the absence of ground\ntruth alignments for supervision, we train a network on the task of shape\nalignment using incomplete shapes generated from full shapes for\nself-supervision. Our network, called ALIGNet, is trained to warp complete\nsource shapes to incomplete targets, as if the target shapes were complete,\nthus essentially rendering the alignment partial-shape agnostic. We aim for the\nnetwork to develop specialized expertise over the common characteristics of the\nshapes in each dataset, thereby achieving a higher-level understanding of the\nexpected shape space to which a local approach would be oblivious. We constrain\nALIGNet through an anisotropic total variation identity regularization to\npromote piecewise smooth deformation fields, facilitating both partial-shape\nagnosticism and post-deformation applications. We demonstrate that ALIGNet\nlearns to align geometrically distinct shapes, and is able to infer plausible\nmappings even when the target shape is significantly incomplete. We show that\nour network learns the common expected characteristics of shape collections,\nwithout over-fitting or memorization, enabling it to produce plausible\ndeformations on unseen data during test time.","url_abs":"http://arxiv.org/abs/1804.08497v2","url_pdf":"http://arxiv.org/pdf/1804.08497v2.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":"alignet-partial-shape-agnostic-alignment-via","repo_url":"https://github.com/ranahanocka/ALIGNet","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":0,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"memorization","task_name":"Memorization"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}