{"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/jsis3d-joint-semantic-instance-segmentation","title":"JSIS3D: Joint Semantic-Instance Segmentation of 3D Point Clouds with Multi-Task Pointwise Networks and Multi-Value Conditional Random Fields","arxiv_id":"1904.00699","date":"2019-04-01","proceeding":"CVPR 2019 6","authors":["Quang-Hieu Pham","Duc Thanh Nguyen","Binh-Son Hua","Gemma Roig","Sai-Kit Yeung"],"abstract":"Deep learning techniques have become the to-go models for most vision-related\ntasks on 2D images. However, their power has not been fully realised on several\ntasks in 3D space, e.g., 3D scene understanding. In this work, we jointly\naddress the problems of semantic and instance segmentation of 3D point clouds.\nSpecifically, we develop a multi-task pointwise network that simultaneously\nperforms two tasks: predicting the semantic classes of 3D points and embedding\nthe points into high-dimensional vectors so that points of the same object\ninstance are represented by similar embeddings. We then propose a multi-value\nconditional random field model to incorporate the semantic and instance labels\nand formulate the problem of semantic and instance segmentation as jointly\noptimising labels in the field model. The proposed method is thoroughly\nevaluated and compared with existing methods on different indoor scene datasets\nincluding S3DIS and SceneNN. Experimental results showed the robustness of the\nproposed joint semantic-instance segmentation scheme over its single\ncomponents. Our method also achieved state-of-the-art performance on semantic\nsegmentation.","url_abs":"http://arxiv.org/abs/1904.00699v2","url_pdf":"http://arxiv.org/pdf/1904.00699v2.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":"jsis3d-joint-semantic-instance-segmentation","repo_url":"https://github.com/pqhieu/JSIS3D","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":null}],"tasks":[{"task_slug":"3d-instance-segmentation-1","task_name":"3D Instance Segmentation"},{"task_slug":"3d-semantic-instance-segmentation","task_name":"3D Semantic Instance Segmentation"},{"task_slug":"3d-semantic-segmentation","task_name":"3D Semantic Segmentation"},{"task_slug":"scene-understanding","task_name":"Scene Understanding"},{"task_slug":"segmentation","task_name":"Segmentation"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-instance-segmentation-on-scenenn-1","task":"3D Instance Segmentation","dataset":"SceneNN","model":"MLS-CRF","rank_in_archive_order":2,"of":3,"metrics":{"mAP@0.5":"12.1"},"uses_additional_data":false},{"leaderboard":"/sota/3d-instance-segmentation-on-scenenn-1","task":"3D Instance Segmentation","dataset":"SceneNN","model":"MT-PNet","rank_in_archive_order":3,"of":3,"metrics":{"mAP@0.5":"8.5"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1904.00699","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}