{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,3,4]],"date-time":"2026-03-04T14:35:42Z","timestamp":1772634942424,"version":"3.50.1"},"reference-count":64,"publisher":"Frontiers Media SA","license":[{"start":{"date-parts":[[2025,9,8]],"date-time":"2025-09-08T00:00:00Z","timestamp":1757289600000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0\/"}],"content-domain":{"domain":["frontiersin.org"],"crossmark-restriction":true},"short-container-title":["Front. Neurorobot."],"abstract":"<jats:sec><jats:title>Introduction<\/jats:title><jats:p>Understanding group actions in real-world settings is essential for the advancement of applications in surveillance, robotics, and autonomous systems. Group activity recognition, particularly in sports scenarios, presents unique challenges due to dynamic interactions, occlusions, and varying viewpoints. To address these challenges, we develop a deep learning system that recognizes multi-person behaviors by integrating appearance-based features (HOG, LBP, SIFT), skeletal data (MediaPipe, MOCON), and motion features. Our approach employs a Dynamic Graph Neural Network (DGNN) and Bi-LSTM architecture, enabling robust recognition of group activities in diverse and dynamic environments. To further validate our framework\u2019s adaptability, we include evaluations on Volleyball and SoccerTrack UAV-recorded datasets, which offer unique perspectives and challenges.<\/jats:p><\/jats:sec><jats:sec><jats:title>Method<\/jats:title><jats:p>Our framework integrates YOLOv11 for object detection and SORT for tracking to extract multi-modal features\u2014including HOG, LBP, SIFT, skeletal data (MediaPipe), and motion context (MOCON). These features are optimized using genetic algorithms and fused within a Dynamic Graph Neural Network (DGNN), which models players as nodes in a spatio-temporal graph, effectively capturing both spatial formations and temporal dynamics.<\/jats:p><\/jats:sec><jats:sec><jats:title>Results<\/jats:title><jats:p>We evaluated our framework on three datasets: a volleyball dataset, SoccerTrack UAV-based soccer dataset, and NBA basketball dataset. Our system achieved 94.5% accuracy on the volleyball dataset (mAP: 94.2%, MPCA: 93.8%) with an inference time of 0.18\u202fs per frame. On the SoccerTrack UAV dataset, accuracy was 91.8% (mAP: 91.5%, MPCA: 90.5%) with 0.20\u202fs inference, and on the NBA basketball dataset, it was 91.1% (mAP: 90.8%, MPCA: 89.8%) with the same 0.20\u202fs per frame. These results highlight our framework\u2019s high performance and efficient computational efficiency across various sports and perspectives.<\/jats:p><\/jats:sec><jats:sec><jats:title>Discussion<\/jats:title><jats:p>Our approach demonstrates robust performance in recognizing multi-person actions across diverse conditions, highlighting its adaptability to both conventional and UAV-based video sources.<\/jats:p><\/jats:sec>","DOI":"10.3389\/fnbot.2025.1631998","type":"journal-article","created":{"date-parts":[[2025,9,8]],"date-time":"2025-09-08T14:16:58Z","timestamp":1757341018000},"update-policy":"https:\/\/doi.org\/10.3389\/crossmark-policy","source":"Crossref","is-referenced-by-count":12,"title":["Dynamic graph neural networks for UAV-based group activity recognition in structured team sports"],"prefix":"10.3389","volume":"19","author":[{"given":"Ishrat","family":"Zahra","sequence":"first","affiliation":[]},{"given":"Yanfeng","family":"Wu","sequence":"additional","affiliation":[]},{"given":"Haifa F.","family":"Alhasson","sequence":"additional","affiliation":[]},{"given":"Shuaa S.","family":"Alharbi","sequence":"additional","affiliation":[]},{"given":"Hanan","family":"Aljuaid","sequence":"additional","affiliation":[]},{"given":"Ahmad","family":"Jalal","sequence":"additional","affiliation":[]},{"given":"Hui","family":"Liu","sequence":"additional","affiliation":[]}],"member":"1965","published-online":{"date-parts":[[2025,9,8]]},"reference":[{"key":"ref1","doi-asserted-by":"publisher","first-page":"1443678","DOI":"10.3389\/fnbot.2024.1443678","article-title":"Unmanned aerial vehicles for human detection and recognition using neural-network model","volume":"18","author":"Abbas","year":"2024","journal-title":"Front. Neurorobot."},{"key":"ref2","first-page":"5229","article-title":"Self-supervised video interaction classification using image representation of skeleton data","author":"Askari","year":"2023"},{"key":"ref3","article-title":"Social scene understanding: end-to-end multi-person action localization and collective activity recognition","volume-title":"Arxiv","author":"Bagautdinov","year":"2016"},{"key":"ref4","doi-asserted-by":"publisher","first-page":"152009","DOI":"10.1109\/ACCESS.2024.3479988","article-title":"Human detection from unmanned aerial vehicles' images for search and rescue missions: a state-of-the-art review","volume":"12","author":"Bany Abdelnabi","year":"2024","journal-title":"IEEE Access"},{"key":"ref5","doi-asserted-by":"publisher","first-page":"850512","DOI":"10.3389\/fphys.2022.850512","article-title":"Drone-based position detection in sports\u2014validation and applications","volume":"13","author":"Bastida-Castillo","year":"2022","journal-title":"Front. Physiol."},{"key":"ref6","first-page":"278","article-title":"How good is good enough?: the impact of errors in single person action classification on the modeling of group interactions in volleyball","author":"Beenhakker","year":"2020"},{"key":"ref7","first-page":"1625","article-title":"Structural recurrent neural network (SRNN) for group activity analysis","author":"Biswas","year":"2018"},{"key":"ref8","doi-asserted-by":"publisher","first-page":"47","DOI":"10.1007\/s10489-007-0074-y","article-title":"Detecting small group activities from multimodal observations","volume":"30","author":"Brdiczka","year":"2009","journal-title":"Appl. Intell."},{"key":"ref9","first-page":"9686","article-title":"Observation-centric SORT: rethinking SORT for robust multi-object tracking","author":"Cao","year":"2023"},{"key":"ref10","doi-asserted-by":"publisher","first-page":"1905","DOI":"10.1109\/TIP.2015.2409564","article-title":"Learning person-person interaction in collective activity recognition","volume":"24","author":"Chang","year":"2015","journal-title":"IEEE Trans. Image Process."},{"key":"ref11","doi-asserted-by":"publisher","first-page":"4405","DOI":"10.1007\/s11042-015-3177-1","article-title":"A survey of depth and inertial sensor fusion for human action recognition","volume":"76","author":"Chen","year":"2017","journal-title":"Multimed. Tools Appl."},{"key":"ref9003","doi-asserted-by":"publisher","first-page":"4194","DOI":"10.1109\/TVT.2023.3327571","article-title":"Global-and-local attention-based reinforcement learning for cooperative behaviour control of multiple UAVs","volume":"73","author":"Chen","year":"2023","journal-title":"IEEE Trans. Veh. Technol."},{"key":"ref12","doi-asserted-by":"publisher","first-page":"137","DOI":"10.1080\/07038992.2021.2024","article-title":"Detection and tracking of belugas, kayaks and motorized boats in drone video using deep learning","volume":"48","author":"Dawson","year":"2022","journal-title":"Can. J. Remote. Sens."},{"key":"ref13","article-title":"Group activity detection from trajectory and video data in soccer","volume-title":"Arxiv","author":"Deli\u00e8ge","year":"2020"},{"key":"ref14","doi-asserted-by":"publisher","first-page":"69","DOI":"10.1007\/978-3-642-33786-4_6","article-title":"Team activity recognition in sports","volume":"7578","author":"Direko\u011flu","year":"2012","journal-title":"Comput. Vis."},{"key":"ref15","doi-asserted-by":"crossref","DOI":"10.1007\/978-3-030-58545-7_11","article-title":"Joint learning of social groups, individuals action and sub-group activities in videos","volume-title":"Arxiv","author":"Ehsanpour","year":"2020"},{"key":"ref16","doi-asserted-by":"publisher","first-page":"1","DOI":"10.3390\/fi12080133","article-title":"Multimodal deep learning for group activity recognition in smart office environments","volume":"12","author":"Florea","year":"2020","journal-title":"Future Internet"},{"key":"ref17","first-page":"836","article-title":"Actor-transformers for group activity recognition","author":"Gavrilyuk","year":"2020"},{"key":"ref18","first-page":"2990","article-title":"Dual-AI: dual-path actor interaction learning for group activity recognition","author":"Han","year":"2022"},{"key":"ref19","doi-asserted-by":"publisher","first-page":"1013","DOI":"10.3233\/JSA-220649","article-title":"Group activity recognition in basketball tracking data \u2013 neural embeddings in team sports (NETS)","volume":"8","author":"Hauri","year":"2022","journal-title":"J. Sports Anal."},{"key":"ref20","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/TIM.2023.3246469","article-title":"Two-domain joint attention mechanism based on sensor data for group activity recognition","volume":"72","author":"Huan","year":"2023","journal-title":"IEEE Trans. Emerg. Top. Comput. Intell."},{"key":"ref21","first-page":"1971","article-title":"A hierarchical deep temporal model for group activity recognition","author":"Ibrahim","year":"2016"},{"key":"ref22","doi-asserted-by":"publisher","first-page":"104298","DOI":"10.1016\/j.jvcir.2024.104298","article-title":"Diving deep into human action recognition in aerial videos: a survey","volume":"104","author":"Kapoor","year":"2024","journal-title":"J. Vis. Commun. Image Represent."},{"key":"ref23","doi-asserted-by":"publisher","first-page":"1678","DOI":"10.1007\/s10618-017-0495-0","article-title":"Activity recognition in beach volleyball using a deep convolutional neural network","volume":"31","author":"Kautz","year":"2017","journal-title":"Data Min. Knowl. Disc."},{"key":"ref24","article-title":"YOLOv11: an overview of the key architectural enhancements","volume-title":"Arxiv","author":"Khanam","year":"2024"},{"key":"ref25","first-page":"13648","article-title":"GroupFormer: group activity recognition with clustered spatial-temporal transformer","author":"Li","year":"2021"},{"key":"ref26","first-page":"2450","article-title":"Learning multi-modal densities on discriminative temporal interaction manifold for group activity recognition","author":"Li","year":"2009"},{"key":"ref27","first-page":"2895","article-title":"SBGAR: semantics based group activity recognition","author":"Li","year":"2017"},{"key":"ref28","first-page":"1059","article-title":"Region-based activity recognition using conditional CAN","author":"Li","year":"2017"},{"key":"ref29","doi-asserted-by":"publisher","first-page":"2785","DOI":"10.1007\/s11263-022-01576-4","article-title":"Transformer-based action recognition for surveillance systems: a case study in crowd analysis","volume":"130","author":"Li","year":"2022","journal-title":"Int. J. Comput. Vis."},{"key":"ref30","doi-asserted-by":"publisher","first-page":"2684","DOI":"10.1109\/TPAMI.2019.2916873","article-title":"NTU RGB+D 120: a large-scale benchmark for 3D human activity understanding","volume":"42","author":"Liu","year":"2020","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"ref31","doi-asserted-by":"publisher","first-page":"195","DOI":"10.1016\/j.neucom.2018.09.060","article-title":"A two-level attention-based interaction model for multi-person activity recognition","volume":"322","author":"Lu","year":"2018","journal-title":"Neurocomputing"},{"key":"ref9002","first-page":"25","article-title":"Comprehensive analysis of human action recognition and object detection in aerial environments","volume-title":"Computer Vision for UAVs: Advanced Methods and Applications","author":"Maheriya","year":"2023"},{"key":"ref9001","first-page":"461","article-title":"Vision Transformer in Sports Action: Recognizing Athletic Activities Across Varied Sporting Domains","volume-title":"In Proceedings of International Conference on Data Science and Applications","author":"Maheriya","year":"2025"},{"key":"ref32","doi-asserted-by":"publisher","first-page":"29","DOI":"10.1109\/ICST.2024.9876543","article-title":"Real-time action recognition in sports: improving recognition accuracy in soccer using CNN-based models","volume":"2024","author":"Maheriya","year":"2024","journal-title":"Proc. Int. Conf. Sports Technol."},{"key":"ref33","doi-asserted-by":"crossref","first-page":"43","DOI":"10.1007\/978-3-319-93025-1_4","article-title":"Genetic algorithm","volume-title":"Evolutionary algorithms and neural networks: Theory and applications","author":"Mirjalili","year":"2019"},{"key":"ref34","doi-asserted-by":"publisher","first-page":"7299","DOI":"10.3390\/s20247299","article-title":"Histogram of oriented gradient-based fusion of features for human action recognition in action video sequences","volume":"20","author":"Patel","year":"2020","journal-title":"Sensors"},{"key":"ref9005","doi-asserted-by":"publisher","first-page":"1369","DOI":"10.1109\/I3CEET61722.2024.10993810","article-title":"Drones: categories, programs, and technological challenges","author":"Patel","year":"2024","journal-title":"In Proceedings of the 2024 International Conference on Communication, Computing and Energy Efficient Technologies (I3CEET)"},{"key":"ref35","doi-asserted-by":"publisher","first-page":"366","DOI":"10.1109\/TMM.2021.3050642","article-title":"Interaction relational network for mutual action recognition","volume":"24","author":"Perez","year":"2022","journal-title":"IEEE Trans. Multi."},{"key":"ref9007","doi-asserted-by":"publisher","first-page":"612","DOI":"10.1080\/02640414.2024.1234567","article-title":"NBA tracking and analytics: trends in team adoption and performance impact","volume":"42","author":"Pr\u00fc\u00dfner","year":"2024","journal-title":"J. Sports Sci."},{"key":"ref36","first-page":"104","article-title":"stagNet: an attentive semantic RNN for group activity recognition","volume-title":"Lecture notes in computer science","author":"Qi","year":"2018"},{"key":"ref9004","doi-asserted-by":"publisher","first-page":"777","DOI":"10.3390\/s24030777","article-title":"Real-time object detection for autonomous solar farm inspection via UAVs","volume":"24","author":"Rodriguez-Vazquez","year":"2024","journal-title":"Sensors"},{"key":"ref37","doi-asserted-by":"publisher","first-page":"423","DOI":"10.5194\/isprs-archives-XLVI-4-W3-2021-423-2022","article-title":"Detecting and recognizing drones using a deep CNN","volume":"46","author":"Samadzadegan","year":"2022","journal-title":"Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci."},{"key":"ref38","doi-asserted-by":"publisher","first-page":"e0132894","DOI":"10.1371\/journal.pone.0132894","article-title":"Exploring game performance in the National Basketball Association Using Player Tracking Data","volume":"10","author":"Sampaio","year":"2015","journal-title":"PLoS One"},{"key":"ref39","first-page":"5083","article-title":"SoccerTrack: a dataset and tracking algorithm for soccer with fish-eye","author":"Scott","year":"2022"},{"key":"ref40","first-page":"7904","article-title":"Skeleton-based action recognition with directed graph neural networks","author":"Shi","year":"2019"},{"key":"ref41","doi-asserted-by":"crossref","DOI":"10.1109\/CVPR.2017.453","article-title":"CERN: confidence-energy recurrent network for group activity recognition","volume-title":"Arxiv","author":"Shu","year":"2017"},{"key":"ref42","doi-asserted-by":"publisher","first-page":"24","DOI":"10.3390\/app10010024","article-title":"Use of machine learning to automate the identification of basketball strategies using whole team player tracking data","volume":"10","author":"Sicilia","year":"2019","journal-title":"Appl. Sci."},{"key":"ref43","doi-asserted-by":"publisher","first-page":"79143","DOI":"10.1109\/ACCESS.2021.3082932","article-title":"Foundations and modeling of dynamic networks using dynamic graph neural networks: a survey","volume":"9","author":"Skarding","year":"2021","journal-title":"IEEE Access"},{"key":"ref44","doi-asserted-by":"publisher","first-page":"4269","DOI":"10.1007\/s11263-024-02082-y","article-title":"Design and analysis of efficient attention in transformers for social group activity recognition","volume":"132","author":"Tamura","year":"2024","journal-title":"Int. J. Comput. Vis."},{"key":"ref45","first-page":"1769","article-title":"GrabCut in one cut","author":"Tang","year":"2013"},{"key":"ref46","first-page":"1283","article-title":"Mining semantics-preserving attention for group activity recognition","author":"Tang","year":"2018"},{"key":"ref47","doi-asserted-by":"crossref","first-page":"85","DOI":"10.1007\/978-3-031-76197-3_8","article-title":"Skeleton-based posture estimation for human action recognition using deep learning","volume-title":"Computational intelligence methods for green technology and sustainable development","author":"Truong","year":"2024"},{"key":"ref48","article-title":"Group activity recognition in computer vision: a comprehensive review, challenges, and future perspectives","volume-title":"Arxiv","author":"Wang","year":"2023"},{"key":"ref49","doi-asserted-by":"publisher","first-page":"53880","DOI":"10.1109\/ACCESS.2023.3282311","article-title":"A comprehensive survey of RGB-based and skeleton-based human action recognition","volume":"11","author":"Wang","year":"2023","journal-title":"IEEE Access"},{"key":"ref9006","doi-asserted-by":"publisher","first-page":"1","DOI":"10.1109\/EMBC53108.2024.10781770","article-title":"mmYOLOH-p: a clinically-oriented mmWave-based human pose estimation tool for unobtrusive patient monitoring","author":"Williams","year":"2024","journal-title":"In Proceedings of the 46th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC)"},{"key":"ref50","first-page":"9956","article-title":"Learning actor relation graphs for group activity recognition","author":"Wu","year":"2019"},{"key":"ref51","doi-asserted-by":"publisher","first-page":"656","DOI":"10.1371\/journal.pone.0319656","article-title":"Maf-net: a multimodal data fusion approach for human action recognition","volume":"20","author":"Xie","year":"2025","journal-title":"PLoS One"},{"key":"ref52","doi-asserted-by":"publisher","first-page":"57","DOI":"10.1016\/j.neunet.2022.12.005","article-title":"DroneAttention: sparse weighted temporal attention for drone-camera based activity recognition","volume":"159","author":"Yadav","year":"2023","journal-title":"Neural Netw."},{"key":"ref53","first-page":"1292","article-title":"Participation-contributed temporal dynamic model for group activity recognition","author":"Yan","year":"2018"},{"key":"ref54","doi-asserted-by":"publisher","first-page":"7444","DOI":"10.1609\/aaai.v32i1.12328","article-title":"Spatial temporal graph convolutional networks for skeleton-based action recognition","volume":"32","author":"Yan","year":"2018","journal-title":"Proc. AAAI Conf. Artif. Intell."},{"key":"ref55","first-page":"10412","article-title":"Learning group activities from skeletons without individual action labels","author":"Zappardino","year":"2021"},{"key":"ref56","article-title":"Group activity recognition via dynamic composition and interaction","volume-title":"Arxiv","author":"Zhang","year":"2023"},{"key":"ref57","doi-asserted-by":"publisher","first-page":"13","DOI":"10.3390\/info14010013","article-title":"Basketball action recognition method of deep neural network based on dynamic residual attention mechanism","volume":"14","author":"Zhang","year":"2023","journal-title":"Information"}],"container-title":["Frontiers in Neurorobotics"],"original-title":[],"link":[{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fnbot.2025.1631998\/full","content-type":"unspecified","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2025,9,8]],"date-time":"2025-09-08T14:17:13Z","timestamp":1757341033000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.frontiersin.org\/articles\/10.3389\/fnbot.2025.1631998\/full"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2025,9,8]]},"references-count":64,"alternative-id":["10.3389\/fnbot.2025.1631998"],"URL":"https:\/\/doi.org\/10.3389\/fnbot.2025.1631998","relation":{},"ISSN":["1662-5218"],"issn-type":[{"value":"1662-5218","type":"electronic"}],"subject":[],"published":{"date-parts":[[2025,9,8]]},"article-number":"1631998"}}