{"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/o-cnn-octree-based-convolutional-neural","title":"O-CNN: Octree-based Convolutional Neural Networks for 3D Shape Analysis","arxiv_id":"1712.01537","date":"2017-12-05","proceeding":null,"authors":["Peng-Shuai Wang","Yang Liu","Yu-Xiao Guo","Chun-Yu Sun","Xin Tong"],"abstract":"We present O-CNN, an Octree-based Convolutional Neural Network (CNN) for 3D\nshape analysis. Built upon the octree representation of 3D shapes, our method\ntakes the average normal vectors of a 3D model sampled in the finest leaf\noctants as input and performs 3D CNN operations on the octants occupied by the\n3D shape surface. We design a novel octree data structure to efficiently store\nthe octant information and CNN features into the graphics memory and execute\nthe entire O-CNN training and evaluation on the GPU. O-CNN supports various CNN\nstructures and works for 3D shapes in different representations. By restraining\nthe computations on the octants occupied by 3D surfaces, the memory and\ncomputational costs of the O-CNN grow quadratically as the depth of the octree\nincreases, which makes the 3D CNN feasible for high-resolution 3D models. We\ncompare the performance of the O-CNN with other existing 3D CNN solutions and\ndemonstrate the efficiency and efficacy of O-CNN in three shape analysis tasks,\nincluding object classification, shape retrieval, and shape segmentation.","url_abs":"http://arxiv.org/abs/1712.01537v1","url_pdf":"http://arxiv.org/pdf/1712.01537v1.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":"o-cnn-octree-based-convolutional-neural","repo_url":"https://github.com/Microsoft/O-CNN","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"3d-object-classification","task_name":"3D Object Classification"},{"task_slug":null,"task_name":"GPU"},{"task_slug":"retrieval","task_name":"Retrieval"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-object-classification-on-modelnet40","task":"3D Object Classification","dataset":"ModelNet40","model":"O-CNN(6)","rank_in_archive_order":4,"of":7,"metrics":{"Classification Accuracy":"89.9"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-scannet","task":"Semantic Segmentation","dataset":"ScanNet","model":"O-CNN","rank_in_archive_order":24,"of":45,"metrics":{"test mIoU":"76.2","val mIoU":"74.0"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}