{"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/general-purpose-deep-point-cloud-feature","title":"General-Purpose Deep Point Cloud Feature Extractor","arxiv_id":null,"date":"2018-03-12","proceeding":"IEEE Winter Conference on Applications of Computer Vision (WACV) 2018 3","authors":["Miguel Dominguez","Rohan Dhamdhere","Atir Petkar","Saloni Jain","Shagan Sah","Raymond Ptucha"],"abstract":"Depth sensors used in autonomous driving and gaming\r\nsystems often report back 3D point clouds. The lack of\r\nstructure from these sensors does not allow these systems\r\nto take advantage of recent advances in convolutional neural networks which are dependent upon traditional filtering\r\nand pooling operations. Analogous to image based convolutional architectures, recently introduced graph based architectures afford similar filtering and pooling operations\r\non arbitrary graphs. We adopt these graph based methods\r\nto 3D point clouds to introduce a generic vector representation of 3D graphs, we call graph 3D (G3D). We believe\r\nwe are the first to use large scale transfer learning on 3D\r\npoint cloud data and demonstrate the discriminant power\r\nof our salient latent representation of 3D point clouds on\r\nunforeseen test sets. By using our G3D network (G3DNet)\r\nas a feature extractor, and then pairing G3D feature vectors\r\nwith a standard classifier, we achieve the best accuracy on\r\nModelNet10 (93.1%) and ModelNet 40 (91.7%) for a graph\r\nnetwork, and comparable performance on the Sydney Urban Objects dataset to other methods. This general-purpose\r\nfeature extractor can be used as an off-the-shelf component\r\nin other 3D scene understanding or object tracking works.","url_abs":"https://ieeexplore.ieee.org/abstract/document/8354322","url_pdf":"https://ieeexplore.ieee.org/stamp/stamp.jsp?tp=&arnumber=8354322","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":"general-purpose-deep-point-cloud-feature","repo_url":"https://github.com/WDot/G3DNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"tf","reach":null}],"tasks":[{"task_slug":"3d-object-classification","task_name":"3D Object Classification"},{"task_slug":"3d-point-cloud-classification","task_name":"3D Point Cloud Classification"},{"task_slug":"autonomous-driving","task_name":"Autonomous Driving"},{"task_slug":"object-tracking","task_name":"Object Tracking"},{"task_slug":"scene-understanding","task_name":"Scene Understanding"},{"task_slug":"transfer-learning","task_name":"Transfer Learning"}],"methods":[{"method_slug":"graph-convolutional-networks","method_name":"Graph Convolutional Networks"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-object-classification-on-modelnet10","task":"3D Object Classification","dataset":"ModelNet10","model":"G3DNet-18 SVM, Fine-Tuned, Vote","rank_in_archive_order":3,"of":4,"metrics":{"Accuracy":"93.1"},"uses_additional_data":true},{"leaderboard":"/sota/3d-object-classification-on-modelnet40","task":"3D Object Classification","dataset":"ModelNet40","model":"G3DNet-18 MLP, Fine-Tuned, Vote","rank_in_archive_order":2,"of":7,"metrics":{"Classification Accuracy":"91.7"},"uses_additional_data":true},{"leaderboard":"/sota/3d-point-cloud-classification-on-modelnet40","task":"3D Point Cloud Classification","dataset":"ModelNet40","model":"G3DNet-18 MLP, Fine-Tuned, Vote","rank_in_archive_order":97,"of":111,"metrics":{"Overall Accuracy":"91.7"},"uses_additional_data":true},{"leaderboard":"/sota/3d-point-cloud-classification-on-sydney-urban","task":"3D Point Cloud Classification","dataset":"Sydney Urban Objects","model":"G3DNet-18 MLP Fine-Tuned, Vote","rank_in_archive_order":3,"of":3,"metrics":{"F1":"72.7"},"uses_additional_data":false}],"syntology":{"atlas_url":null,"mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}