{"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/foldingnet-point-cloud-auto-encoder-via-deep","title":"FoldingNet: Point Cloud Auto-encoder via Deep Grid Deformation","arxiv_id":"1712.07262","date":"2017-12-19","proceeding":"CVPR 2018 6","authors":["Yaoqing Yang","Chen Feng","Yiru Shen","Dong Tian"],"abstract":"Recent deep networks that directly handle points in a point set, e.g.,\nPointNet, have been state-of-the-art for supervised learning tasks on point\nclouds such as classification and segmentation. In this work, a novel\nend-to-end deep auto-encoder is proposed to address unsupervised learning\nchallenges on point clouds. On the encoder side, a graph-based enhancement is\nenforced to promote local structures on top of PointNet. Then, a novel\nfolding-based decoder deforms a canonical 2D grid onto the underlying 3D object\nsurface of a point cloud, achieving low reconstruction errors even for objects\nwith delicate structures. The proposed decoder only uses about 7% parameters of\na decoder with fully-connected neural networks, yet leads to a more\ndiscriminative representation that achieves higher linear SVM classification\naccuracy than the benchmark. In addition, the proposed decoder structure is\nshown, in theory, to be a generic architecture that is able to reconstruct an\narbitrary point cloud from a 2D grid. Our code is available at\nhttp://www.merl.com/research/license#FoldingNet","url_abs":"http://arxiv.org/abs/1712.07262v2","url_pdf":"http://arxiv.org/pdf/1712.07262v2.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":"foldingnet-point-cloud-auto-encoder-via-deep","repo_url":"https://github.com/AnTao97/UnsupervisedPointCloudReconstruction","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}},{"paper_slug":"foldingnet-point-cloud-auto-encoder-via-deep","repo_url":"https://github.com/XuyangBai/FoldingNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null},{"paper_slug":"foldingnet-point-cloud-auto-encoder-via-deep","repo_url":"https://github.com/qinglew/FoldingNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"3d-point-cloud-linear-classification","task_name":"3D Point Cloud Linear Classification"},{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"classification","task_name":"General Classification"},{"task_slug":"unsupervised-3d-point-cloud-linear-evaluation","task_name":"Unsupervised 3D Point Cloud Linear Evaluation"}],"methods":[{"method_slug":"svm","method_name":"SVM"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-point-cloud-linear-classification-on","task":"3D Point Cloud Linear Classification","dataset":"ModelNet40","model":"FoldingNet","rank_in_archive_order":17,"of":20,"metrics":{"Overall Accuracy":"88.4"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1712.07262","atlas_url":"https://app.syntology.ai/?focus=1712.07262","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}