{"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/point2sequence-learning-the-shape","title":"Point2Sequence: Learning the Shape Representation of 3D Point Clouds with an Attention-based Sequence to Sequence Network","arxiv_id":"1811.02565","date":"2018-11-06","proceeding":null,"authors":["Xinhai Liu","Zhizhong Han","Yu-Shen Liu","Matthias Zwicker"],"abstract":"Exploring contextual information in the local region is important for shape\nunderstanding and analysis. Existing studies often employ hand-crafted or\nexplicit ways to encode contextual information of local regions. However, it is\nhard to capture fine-grained contextual information in hand-crafted or explicit\nmanners, such as the correlation between different areas in a local region,\nwhich limits the discriminative ability of learned features. To resolve this\nissue, we propose a novel deep learning model for 3D point clouds, named\nPoint2Sequence, to learn 3D shape features by capturing fine-grained contextual\ninformation in a novel implicit way. Point2Sequence employs a novel sequence\nlearning model for point clouds to capture the correlations by aggregating\nmulti-scale areas of each local region with attention. Specifically,\nPoint2Sequence first learns the feature of each area scale in a local region.\nThen, it captures the correlation between area scales in the process of\naggregating all area scales using a recurrent neural network (RNN) based\nencoder-decoder structure, where an attention mechanism is proposed to\nhighlight the importance of different area scales. Experimental results show\nthat Point2Sequence achieves state-of-the-art performance in shape\nclassification and segmentation tasks.","url_abs":"http://arxiv.org/abs/1811.02565v2","url_pdf":"http://arxiv.org/pdf/1811.02565v2.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":"3d-part-segmentation","task_name":"3D Part Segmentation"},{"task_slug":"3d-point-cloud-classification","task_name":"3D Point Cloud Classification"},{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"shape-representation-of-3d-point-clouds","task_name":"Shape Representation Of 3D Point Clouds"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/3d-part-segmentation-on-shapenet-part","task":"3D Part Segmentation","dataset":"ShapeNet-Part","model":"P2Sequence","rank_in_archive_order":52,"of":67,"metrics":{"Instance Average IoU":"85.2"},"uses_additional_data":false},{"leaderboard":"/sota/3d-point-cloud-classification-on-modelnet40","task":"3D Point Cloud Classification","dataset":"ModelNet40","model":"P2Sequence","rank_in_archive_order":85,"of":111,"metrics":{"Overall Accuracy":"92.6"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.02565","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}