{"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/generalized-graph-convolutional-networks-for","title":"Optimized Skeleton-based Action Recognition via Sparsified Graph Regression","arxiv_id":"1811.12013","date":"2018-11-29","proceeding":null,"authors":["Xiang Gao","Wei Hu","Jiaxiang Tang","Jiaying Liu","Zongming Guo"],"abstract":"With the prevalence of accessible depth sensors, dynamic human body skeletons\nhave attracted much attention as a robust modality for action recognition.\nPrevious methods model skeletons based on RNN or CNN, which has limited\nexpressive power for irregular skeleton joints. While graph convolutional\nnetworks (GCN) have been proposed to address irregular graph-structured data,\nthe fundamental graph construction remains challenging. In this paper, we\nrepresent skeletons naturally on graphs, and propose a graph regression based\nGCN (GR-GCN) for skeleton-based action recognition, aiming to capture the\nspatio-temporal variation in the data. As the graph representation is crucial\nto graph convolution, we first propose graph regression to statistically learn\nthe underlying graph from multiple observations. In particular, we provide\nspatio-temporal modeling of skeletons and pose an optimization problem on the\ngraph structure over consecutive frames, which enforces the sparsity of the\nunderlying graph for efficient representation. The optimized graph not only\nconnects each joint to its neighboring joints in the same frame strongly or\nweakly, but also links with relevant joints in the previous and subsequent\nframes. We then feed the optimized graph into the GCN along with the\ncoordinates of the skeleton sequence for feature learning, where we deploy\nhigh-order and fast Chebyshev approximation of spectral graph convolution.\nFurther, we provide analysis of the variation characterization by the Chebyshev\napproximation. Experimental results validate the effectiveness of the proposed\ngraph regression and show that the proposed GR-GCN achieves the\nstate-of-the-art performance on the widely used NTU RGB+D, UT-Kinect and SYSU\n3D datasets.","url_abs":"http://arxiv.org/abs/1811.12013v2","url_pdf":"http://arxiv.org/pdf/1811.12013v2.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":"action-recognition-in-videos","task_name":"Action Recognition"},{"task_slug":"graph-regression","task_name":"Graph Regression"},{"task_slug":"skeleton-based-action-recognition","task_name":"Skeleton Based Action Recognition"},{"task_slug":"action-recognition","task_name":"Temporal Action Localization"},{"task_slug":"graph-construction","task_name":"graph construction"},{"task_slug":"regression-1","task_name":"regression"}],"methods":[{"method_slug":"gcn","method_name":"GCN"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/skeleton-based-action-recognition-on-florence","task":"Skeleton Based Action Recognition","dataset":"Florence 3D","model":"Complete GR-GCN","rank_in_archive_order":2,"of":7,"metrics":{"Accuracy":"98.4%"},"uses_additional_data":false},{"leaderboard":"/sota/skeleton-based-action-recognition-on-ntu-rgbd","task":"Skeleton Based Action Recognition","dataset":"NTU RGB+D","model":"GR-GCN","rank_in_archive_order":74,"of":135,"metrics":{"Accuracy (CS)":"87.5","Accuracy (CV)":"94.3"},"uses_additional_data":false},{"leaderboard":"/sota/skeleton-based-action-recognition-on-sysu-3d","task":"Skeleton Based Action Recognition","dataset":"SYSU 3D","model":"Complete GR-GCN","rank_in_archive_order":5,"of":9,"metrics":{"Accuracy":"77.9%"},"uses_additional_data":false},{"leaderboard":"/sota/skeleton-based-action-recognition-on-ut","task":"Skeleton Based Action Recognition","dataset":"UT-Kinect","model":"Complete GR-GCN","rank_in_archive_order":3,"of":7,"metrics":{"Accuracy":"98.5%"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=1811.12013","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}