{"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/splatnet-sparse-lattice-networks-for-point","title":"SPLATNet: Sparse Lattice Networks for Point Cloud Processing","arxiv_id":"1802.08275","date":"2018-02-22","proceeding":"CVPR 2018 6","authors":["Hang Su","Varun Jampani","Deqing Sun","Subhransu Maji","Evangelos Kalogerakis","Ming-Hsuan Yang","Jan Kautz"],"abstract":"We present a network architecture for processing point clouds that directly\noperates on a collection of points represented as a sparse set of samples in a\nhigh-dimensional lattice. Naively applying convolutions on this lattice scales\npoorly, both in terms of memory and computational cost, as the size of the\nlattice increases. Instead, our network uses sparse bilateral convolutional\nlayers as building blocks. These layers maintain efficiency by using indexing\nstructures to apply convolutions only on occupied parts of the lattice, and\nallow flexible specifications of the lattice structure enabling hierarchical\nand spatially-aware feature learning, as well as joint 2D-3D reasoning. Both\npoint-based and image-based representations can be easily incorporated in a\nnetwork with such layers and the resulting model can be trained in an\nend-to-end manner. We present results on 3D segmentation tasks where our\napproach outperforms existing state-of-the-art techniques.","url_abs":"http://arxiv.org/abs/1802.08275v4","url_pdf":"http://arxiv.org/pdf/1802.08275v4.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":"splatnet-sparse-lattice-networks-for-point","repo_url":"https://github.com/IsaacRe/splatnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":{"status":"ok","spdx":"NOASSERTION"}},{"paper_slug":"splatnet-sparse-lattice-networks-for-point","repo_url":"https://github.com/NVlabs/splatnet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"caffe2","reach":{"status":"ok","spdx":"NOASSERTION"}}],"tasks":[{"task_slug":"3d-part-segmentation","task_name":"3D Part Segmentation"},{"task_slug":"3d-semantic-segmentation","task_name":"3D Semantic Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-part-segmentation-on-shapenet-part","task":"3D Part Segmentation","dataset":"ShapeNet-Part","model":"SPLATNet 3D","rank_in_archive_order":59,"of":67,"metrics":{"Class Average IoU":"82.0","Instance Average IoU":"84.6"},"uses_additional_data":false},{"leaderboard":"/sota/3d-semantic-segmentation-on-semantickitti","task":"3D Semantic Segmentation","dataset":"SemanticKITTI","model":"SPLATNet","rank_in_archive_order":39,"of":45,"metrics":{"test mIoU":"18.4%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-scannet","task":"Semantic Segmentation","dataset":"ScanNet","model":"SPLAT Net","rank_in_archive_order":44,"of":45,"metrics":{"test mIoU":"39.3"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1802.08275","atlas_url":"https://app.syntology.ai/?focus=1802.08275","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}