{"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/segcloud-semantic-segmentation-of-3d-point","title":"SEGCloud: Semantic Segmentation of 3D Point Clouds","arxiv_id":"1710.07563","date":"2017-10-20","proceeding":null,"authors":["Lyne P. Tchapmi","Christopher B. Choy","Iro Armeni","JunYoung Gwak","Silvio Savarese"],"abstract":"3D semantic scene labeling is fundamental to agents operating in the real\nworld. In particular, labeling raw 3D point sets from sensors provides\nfine-grained semantics. Recent works leverage the capabilities of Neural\nNetworks (NNs), but are limited to coarse voxel predictions and do not\nexplicitly enforce global consistency. We present SEGCloud, an end-to-end\nframework to obtain 3D point-level segmentation that combines the advantages of\nNNs, trilinear interpolation(TI) and fully connected Conditional Random Fields\n(FC-CRF). Coarse voxel predictions from a 3D Fully Convolutional NN are\ntransferred back to the raw 3D points via trilinear interpolation. Then the\nFC-CRF enforces global consistency and provides fine-grained semantics on the\npoints. We implement the latter as a differentiable Recurrent NN to allow joint\noptimization. We evaluate the framework on two indoor and two outdoor 3D\ndatasets (NYU V2, S3DIS, KITTI, Semantic3D.net), and show performance\ncomparable or superior to the state-of-the-art on all datasets.","url_abs":"http://arxiv.org/abs/1710.07563v1","url_pdf":"http://arxiv.org/pdf/1710.07563v1.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":[],"tasks":[{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/semantic-segmentation-on-s3dis-area5","task":"Semantic Segmentation","dataset":"S3DIS Area5","model":"SegCloud","rank_in_archive_order":55,"of":61,"metrics":{"Number of params":"N/A","mAcc":"57.4","mIoU":"48.9"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-semantic3d","task":"Semantic Segmentation","dataset":"Semantic3D","model":"SegCloud","rank_in_archive_order":13,"of":17,"metrics":{"mIoU":"61.3%"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-semantic3d","task":"Semantic Segmentation","dataset":"Semantic3D","model":"3D-FCNN-TI","rank_in_archive_order":16,"of":17,"metrics":{"mIoU":"58.2%"},"uses_additional_data":false}],"syntology":{"syntology_url":"https://syntology.ai/paper/1710.07563","atlas_url":"https://app.syntology.ai/?focus=1710.07563","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}