{"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/scalable-3d-panoptic-segmentation-with","title":"Scalable 3D Panoptic Segmentation As Superpoint Graph Clustering","arxiv_id":"2401.06704","date":"2024-01-12","proceeding":null,"authors":["Damien Robert","Hugo Raguet","Loic Landrieu"],"abstract":"We introduce a highly efficient method for panoptic segmentation of large 3D point clouds by redefining this task as a scalable graph clustering problem. This approach can be trained using only local auxiliary tasks, thereby eliminating the resource-intensive instance-matching step during training. Moreover, our formulation can easily be adapted to the superpoint paradigm, further increasing its efficiency. This allows our model to process scenes with millions of points and thousands of objects in a single inference. Our method, called SuperCluster, achieves a new state-of-the-art panoptic segmentation performance for two indoor scanning datasets: $50.1$ PQ ($+7.8$) for S3DIS Area~5, and $58.7$ PQ ($+25.2$) for ScanNetV2. We also set the first state-of-the-art for two large-scale mobile mapping benchmarks: KITTI-360 and DALES. With only $209$k parameters, our model is over $30$ times smaller than the best-competing method and trains up to $15$ times faster. Our code and pretrained models are available at https://github.com/drprojects/superpoint_transformer.","url_abs":"https://arxiv.org/abs/2401.06704v2","url_pdf":"https://arxiv.org/pdf/2401.06704v2.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":"scalable-3d-panoptic-segmentation-with","repo_url":"https://github.com/drprojects/superpoint_transformer","is_official":1,"mentioned_in_paper":1,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"3d-panoptic-segmentation","task_name":"3D Panoptic Segmentation"},{"task_slug":"3d-semantic-segmentation","task_name":"3D Semantic Segmentation"},{"task_slug":"graph-clustering","task_name":"Graph Clustering"},{"task_slug":"panoptic-segmentation","task_name":"Panoptic Segmentation"},{"task_slug":"segmentation","task_name":"Segmentation"},{"task_slug":"semantic-segmentation","task_name":"Semantic Segmentation"}],"methods":[{"method_slug":"set","method_name":"SET"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-semantic-segmentation-on-dales","task":"3D Semantic Segmentation","dataset":"DALES","model":"SuperCluster","rank_in_archive_order":4,"of":9,"metrics":{"Model size":"210M","mIoU":"77.3"},"uses_additional_data":false},{"leaderboard":"/sota/3d-semantic-segmentation-on-kitti-360","task":"3D Semantic Segmentation","dataset":"KITTI-360","model":"SuperCluster","rank_in_archive_order":8,"of":8,"metrics":{"Model size":"790K","miou Val":"62.1"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-dales","task":"Panoptic Segmentation","dataset":"DALES","model":"SuperCluster","rank_in_archive_order":1,"of":1,"metrics":{"PQ":"61.2","Params (M)":"0.21","RQ":"68.6","SQ":"87.1"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-kitti-360","task":"Panoptic Segmentation","dataset":"KITTI-360","model":"SuperCluster","rank_in_archive_order":1,"of":1,"metrics":{"PQ":"48.3","Params (M)":"0.79","RQ":"58.4","SQ":"75.1"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-s3dis","task":"Panoptic Segmentation","dataset":"S3DIS","model":"SuperCluster","rank_in_archive_order":1,"of":1,"metrics":{"PQ":"55.9","PQ (with stuff)":"62.7","Params (M)":"0.21","RQ":"66.3","RQ (with stuff)":"73.2","SQ":"83.8","SQ (with stuff)":"84.8"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-s3dis-area5","task":"Panoptic Segmentation","dataset":"S3DIS Area5","model":"SuperCluster","rank_in_archive_order":1,"of":5,"metrics":{"PQ":"50.1","PQ (with stuff)":"58.4","Params (M)":"0.21","RQ":"60.1","RQ (with stuff)":"68.4","SQ":"76.6","SQ (with stuff)":"77.8"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-scannet","task":"Panoptic Segmentation","dataset":"ScanNet","model":"SuperCluster","rank_in_archive_order":2,"of":4,"metrics":{"PQ":"58.7","PQ_st":"84.1","PQ_th":"69.1"},"uses_additional_data":false},{"leaderboard":"/sota/panoptic-segmentation-on-scannetv2","task":"Panoptic Segmentation","dataset":"ScanNetV2","model":"SuperCluster","rank_in_archive_order":3,"of":5,"metrics":{"PQ":"58.7","Params (M)":"1","RQ":"69.1","SQ":"84.1"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-s3dis","task":"Semantic Segmentation","dataset":"S3DIS","model":"SuperCluster","rank_in_archive_order":11,"of":54,"metrics":{"Mean IoU":"75.3","Number of params":"0.21M","Params (M)":"0.21"},"uses_additional_data":false},{"leaderboard":"/sota/semantic-segmentation-on-s3dis-area5","task":"Semantic Segmentation","dataset":"S3DIS Area5","model":"SuperCluster","rank_in_archive_order":38,"of":61,"metrics":{"Number of params":"0.21","mIoU":"68.1"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2401.06704","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}