{"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/3d-shapenets-a-deep-representation-for","title":"3D ShapeNets: A Deep Representation for Volumetric Shapes","arxiv_id":"1406.5670","date":"2014-06-22","proceeding":"CVPR 2015 6","authors":["Zhirong Wu","Shuran Song","Aditya Khosla","Fisher Yu","Linguang Zhang","Xiaoou Tang","Jianxiong Xiao"],"abstract":"3D shape is a crucial but heavily underutilized cue in today's computer\nvision systems, mostly due to the lack of a good generic shape representation.\nWith the recent availability of inexpensive 2.5D depth sensors (e.g. Microsoft\nKinect), it is becoming increasingly important to have a powerful 3D shape\nrepresentation in the loop. Apart from category recognition, recovering full 3D\nshapes from view-based 2.5D depth maps is also a critical part of visual\nunderstanding. To this end, we propose to represent a geometric 3D shape as a\nprobability distribution of binary variables on a 3D voxel grid, using a\nConvolutional Deep Belief Network. Our model, 3D ShapeNets, learns the\ndistribution of complex 3D shapes across different object categories and\narbitrary poses from raw CAD data, and discovers hierarchical compositional\npart representations automatically. It naturally supports joint object\nrecognition and shape completion from 2.5D depth maps, and it enables active\nobject recognition through view planning. To train our 3D deep learning model,\nwe construct ModelNet -- a large-scale 3D CAD model dataset. Extensive\nexperiments show that our 3D deep representation enables significant\nperformance improvement over the-state-of-the-arts in a variety of tasks.","url_abs":"http://arxiv.org/abs/1406.5670v3","url_pdf":"http://arxiv.org/pdf/1406.5670v3.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":"3d-shapenets-a-deep-representation-for","repo_url":"https://github.com/AliBahri94/SVWA_TTA","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"3d-shapenets-a-deep-representation-for","repo_url":"https://github.com/hamidreza-dastmalchi/3dd-tta","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}},{"paper_slug":"3d-shapenets-a-deep-representation-for","repo_url":"https://github.com/zhirongw/3DShapeNets","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"none","reach":null}],"tasks":[{"task_slug":"3d-point-cloud-classification","task_name":"3D Point Cloud Classification"},{"task_slug":"3d-shape-representation","task_name":"3D Shape Representation"},{"task_slug":"object","task_name":"Object"},{"task_slug":"object-recognition","task_name":"Object Recognition"}],"methods":[{"method_slug":"deep-belief-network","method_name":"Deep Belief Network"}],"datasets_introduced":[{"slug":"modelnet","name":"ModelNet","full_name":""}],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-point-cloud-classification-on-modelnet40","task":"3D Point Cloud Classification","dataset":"ModelNet40","model":"3DShapeNets","rank_in_archive_order":110,"of":111,"metrics":{"Mean Accuracy":"77.3"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1406.5670","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}