{"status":"ok","message-type":"work","message-version":"1.0.0","message":{"indexed":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T14:07:07Z","timestamp":1784297227477,"version":"3.55.0"},"reference-count":29,"publisher":"Springer Science and Business Media LLC","issue":"1","license":[{"start":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T00:00:00Z","timestamp":1784246400000},"content-version":"tdm","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"},{"start":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T00:00:00Z","timestamp":1784246400000},"content-version":"vor","delay-in-days":0,"URL":"https:\/\/creativecommons.org\/licenses\/by\/4.0"}],"funder":[{"name":"Egypt Japan University"}],"content-domain":{"domain":["link.springer.com"],"crossmark-restriction":false},"short-container-title":["Sci Rep"],"abstract":"<jats:title>Abstract<\/jats:title>\n                  <jats:p>The prediction and recognition of unstable human walking patterns are of high importance for active video surveillance, smart environments, and assistive healthcare, particularly for fall detection in the elderly. This research investigates the utility of Gait Energy Images (GEIs) combined with deep learning, vision transformers, and few-shot learning architectures to enhance the classification of stable and unstable pedestrian walking patterns. We evaluate and compare twelve methodologies: six classical or feature-based machine learning models (Linear SVM, HOG + SVM, LBP + RBF-SVM, Random Forest, XGBoost, and an adapted GaitSet baseline), three deep learning models (MobileNet, Vision Transformer (ViT), and YOLOv8-cls), and three episodic few-shot learning techniques (Prototypical, Matching, and Relation Networks) under data-scarcity regimes. To facilitate this evaluation, we introduce the Unstable and Stable Walking Pedestrian (USWP) dataset, constructed by fusing and harmonizing sequences from seven public action recognition databases, containing 3250 unique GEIs with a subject-independent evaluation protocol to prevent identity-based domain leakage. Our experiments demonstrate that the YOLOv8-cls model achieves an overall accuracy of 96.92% (97.14% on Loss-of-Balance anomalies and 94.67% on Active Motion anomalies), significantly outperforming the conventional Linear SVM baseline (75.38%) and MobileNet (91.08%). Conversely, Relation Networks exhibit lower few-shot performance (71.08%) due to optimization complexities in learning similarity metrics from sparse data. Leave-One-Dataset-Out (LODO) cross-validation reveals an average generalization accuracy of 83.28%, indicating significant domain bias across source databases and underscoring that within-dataset evaluations overestimate real-world generalization. Computational complexity analysis shows that MobileNet provides an optimal trade-off for real-time edge deployment (4.2 ms latency), while preprocessing ablation studies demonstrate that integrating the Segment Anything Model (SAM) with MediaPipe-derived Regions of Interest (ROIs) yields a 12.50% absolute improvement in accuracy by eliminating background noise.<\/jats:p>","DOI":"10.1038\/s41598-026-61650-z","type":"journal-article","created":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T07:24:49Z","timestamp":1784273089000},"update-policy":"https:\/\/doi.org\/10.1007\/springer_crossmark_policy","source":"Crossref","is-referenced-by-count":0,"title":["Walking stability prediction for pedestrians using gait energy images and hybrid deep and few-shot learning models"],"prefix":"10.1038","volume":"16","author":[{"ORCID":"https:\/\/orcid.org\/0000-0002-9148-6960","authenticated-orcid":false,"given":"Mahmoud","family":"Taha","sequence":"first","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ahmed","family":"Fares","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Hirozumi","family":"Yamaguchi","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]},{"given":"Ahmed B.","family":"Zaky","sequence":"additional","affiliation":[],"role":[{"vocabulary":"crossref","role":"author"}]}],"member":"297","published-online":{"date-parts":[[2026,7,17]]},"reference":[{"key":"61650_CR1","doi-asserted-by":"publisher","first-page":"2259","DOI":"10.1007\/s10462-020-09904-8","volume":"54","author":"P Pareek","year":"2021","unstructured":"Pareek, P. & Thakkar, A. A survey on video-based human action recognition: Recent updates, datasets, challenges, and applications. Artif. Intell. Rev. 54, 2259\u20132322. https:\/\/doi.org\/10.1007\/s10462-020-09904-8 (2021).","journal-title":"Artif. Intell. Rev."},{"key":"61650_CR2","doi-asserted-by":"publisher","first-page":"36372","DOI":"10.1109\/ACCESS.2024.3373199","volume":"12","author":"M Karim","year":"2024","unstructured":"Karim, M. et al. Human action recognition systems: A review of the trends and state-of-the-art. IEEE Access 12, 36372\u201336390. https:\/\/doi.org\/10.1109\/ACCESS.2024.3373199 (2024).","journal-title":"IEEE Access"},{"key":"61650_CR3","doi-asserted-by":"crossref","unstructured":"Sykes, E. R. Next-generation fall detection: harnessing human pose estimation and transformer technology. Health Syst. 1\u201319 (2024).","DOI":"10.1080\/20476965.2024.2395574"},{"key":"61650_CR4","doi-asserted-by":"publisher","first-page":"42395","DOI":"10.1007\/s11042-023-14993-y","volume":"82","author":"AN Zereen","year":"2023","unstructured":"Zereen, A. N. et al. Video analytic system for activity profiling, fall detection, and unstable motion detection. Multimedia Tools Appl. 82, 42395\u201342415 (2023).","journal-title":"Multimedia Tools Appl."},{"key":"61650_CR5","first-page":"51","volume":"1","author":"Y Sun","year":"2024","unstructured":"Sun, Y. & Ortiz, J. Machine learning-driven pedestrian recognition and behavior prediction for enhancing public safety in smart cities. J. Artif. Intell. Inf. 1, 51\u201357 (2024).","journal-title":"J. Artif. Intell. Inf."},{"key":"61650_CR6","doi-asserted-by":"publisher","first-page":"316","DOI":"10.1109\/TPAMI.2006.38","volume":"28","author":"J Han","year":"2006","unstructured":"Han, J. & Bhanu, B. Individual recognition using gait energy image. IEEE Trans. Pattern Anal. Mach. Intell. 28, 316\u2013322. https:\/\/doi.org\/10.1109\/TPAMI.2006.38 (2006).","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"61650_CR7","doi-asserted-by":"crossref","unstructured":"Bashir, K., Xiang, T. & Gong, S. Gait recognition using gait entropy image. In Proceedings of the 3rd International Conference on Crime Detection and Prevention (ICDP), 1\u20136 (IEEE, 2010).","DOI":"10.1049\/ic.2009.0230"},{"key":"61650_CR8","doi-asserted-by":"publisher","first-page":"966","DOI":"10.1109\/TCSVT.2012.2186744","volume":"22","author":"W Kusakunniran","year":"2012","unstructured":"Kusakunniran, W., Wu, Q., Zhang, Y. & Li, H. Gait recognition under various viewing angles based on correlated motion regression. IEEE Trans. Circuits Syst. Video Technol. 22, 966\u2013980 (2012).","journal-title":"IEEE Trans. Circuits Syst. Video Technol."},{"key":"61650_CR9","first-page":"122","volume":"29","author":"S Zheng","year":"2011","unstructured":"Zheng, S., Zhang, Y. & Huang, K. Robust view transformation model for gait recognition. Image Vis. Comput. 29, 122\u2013130 (2011).","journal-title":"Image Vis. Comput."},{"key":"61650_CR10","doi-asserted-by":"crossref","unstructured":"Chao, H., He, Y., Zhang, J. & Feng, J. GaitSet: Regarding gait as a set for cross-view gait recognition. In Proceedings of the AAAI Conference on Artificial Intelligence, 1330\u20131337 (AAAI Press, 2019).","DOI":"10.1609\/aaai.v33i01.33018126"},{"key":"61650_CR11","doi-asserted-by":"publisher","first-page":"1533","DOI":"10.1109\/TIP.2021.3055203","volume":"30","author":"H Chao","year":"2021","unstructured":"Chao, H., Zhang, J., Feng, J. & Wang, J. Gaitset: Dynamic representation learning for gait recognition. IEEE Trans. Image Process. 30, 1533\u20131545. https:\/\/doi.org\/10.1109\/TIP.2021.3055203 (2021).","journal-title":"IEEE Trans. Image Process."},{"key":"61650_CR12","doi-asserted-by":"publisher","unstructured":"Wen, Y. The advancements and future prospects of pedestrian action recognition based on machine learning algorithms. Appl. Comput. Eng. 86, 101\u2013107. https:\/\/doi.org\/10.54254\/2755-2721\/86\/20241566 (2024).","DOI":"10.54254\/2755-2721\/86\/20241566"},{"key":"61650_CR13","doi-asserted-by":"publisher","DOI":"10.1016\/j.knosys.2025.113232","volume":"315","author":"T Sun","year":"2025","unstructured":"Sun, T. et al. Skeletal joint image-based multi-channel fusion network for human activity recognition. Knowl.-Based Syst. 315, 113232. https:\/\/doi.org\/10.1016\/j.knosys.2025.113232 (2025).","journal-title":"Knowl.-Based Syst."},{"key":"61650_CR14","doi-asserted-by":"publisher","unstructured":"Zhao, Y. et al. Multimodal depression assessment framework integrating personality and gait for older adults with medical conditions. IEEE Trans. Affect. Comput. 2048\u20132061. https:\/\/doi.org\/10.1109\/TAFFC.2025.3552835 (2025).","DOI":"10.1109\/TAFFC.2025.3552835"},{"key":"61650_CR15","doi-asserted-by":"publisher","first-page":"2300266","DOI":"10.1002\/aisy.202300266","volume":"5","author":"Y Zhao","year":"2023","unstructured":"Zhao, Y., Sun, T. & Li, W. J. Infusing a convolutional neural network with encoded joint node image data to recognize 25 daily human activities. Adv. Intell. Syst. 5, 2300266. https:\/\/doi.org\/10.1002\/aisy.202300266 (2023).","journal-title":"Adv. Intell. Syst."},{"key":"61650_CR16","doi-asserted-by":"publisher","DOI":"10.1109\/TNNLS.2026.3678220","author":"G Lan","year":"2026","unstructured":"Lan, G., Xiao, S., Wen, J., Yang, J. & Gao, X. A deep neural network optimization framework based on optimal transport bridge feature selection and sparse representation. IEEE Trans. Neural Netw. Learn. Syst. https:\/\/doi.org\/10.1109\/TNNLS.2026.3678220 (2026).","journal-title":"IEEE Trans. Neural Netw. Learn. Syst."},{"key":"61650_CR17","doi-asserted-by":"publisher","DOI":"10.1109\/JBHI.2025.3588196","author":"G Lan","year":"2025","unstructured":"Lan, G., Zhu, Y., Xiao, S., Iqbal, M. & Yang, J. Mitigating data bias in healthcare ai with self-supervised standardization. IEEE J. Biomed. Health Inform. https:\/\/doi.org\/10.1109\/JBHI.2025.3588196 (2025).","journal-title":"IEEE J. Biomed. Health Inform."},{"key":"61650_CR18","doi-asserted-by":"publisher","first-page":"1038","DOI":"10.1109\/tetci.2025.3622647","volume":"10","author":"G Lan","year":"2025","unstructured":"Lan, G. et al. Deep neural networks internal representation via neuron community exploration. IEEE Trans. Emerg. Top. Comput. Intell. 10, 1038\u20131049. https:\/\/doi.org\/10.1109\/tetci.2025.3622647 (2025).","journal-title":"IEEE Trans. Emerg. Top. Comput. Intell."},{"key":"61650_CR19","unstructured":"Lan, G., Xiao, S., Xi, M., Wen, J. & Yang, J. Inner information analysis algorithm for deep neural network based on community. In International Conference on Learning Representations 2025, 4953\u20134973 (2025)."},{"key":"61650_CR20","doi-asserted-by":"publisher","unstructured":"Lan, G., Xiao, S., Yang, J. & Wen, J. Generative model perception rectification algorithm for trade-off between diversity and quality. In Proceedings of the AAAI Conference on Artificial Intelligence 38, 13328\u201313336. https:\/\/doi.org\/10.1609\/aaai.v38i12.29234 (2024).","DOI":"10.1609\/aaai.v38i12.29234"},{"key":"61650_CR21","unstructured":"Dosovitskiy, A. An image is worth 16x16 words: Transformers for image recognition at scale (2020). arXiv:2010.11929."},{"key":"61650_CR22","doi-asserted-by":"publisher","first-page":"87","DOI":"10.1109\/TPAMI.2022.3152247","volume":"45","author":"K Han","year":"2022","unstructured":"Han, K. et al. A survey on vision transformer. IEEE Trans. Pattern Anal. Mach. Intell. 45, 87\u2013110 (2022).","journal-title":"IEEE Trans. Pattern Anal. Mach. Intell."},{"key":"61650_CR23","doi-asserted-by":"publisher","unstructured":"Dahri, F.\u00a0H. et al. Real-time detection of UAV small target\u2019s infrared imaging via ostd yolov8. In Proceedings of IEEE URUCON 2024. https:\/\/doi.org\/10.1109\/URUCON63440.2024.10850067 (2024).","DOI":"10.1109\/URUCON63440.2024.10850067"},{"key":"61650_CR24","doi-asserted-by":"publisher","first-page":"170","DOI":"10.1007\/s44196-025-00911-7","volume":"18","author":"M Javed","year":"2025","unstructured":"Javed, M. et al. Audio-visual synchronization and lip movement analysis for real-time deepfake detection. Int. J. Comput. Intell. Syst. 18, 170 (2025).","journal-title":"Int. J. Comput. Intell. Syst."},{"key":"61650_CR25","doi-asserted-by":"publisher","first-page":"400","DOI":"10.1007\/s11760-025-03970-7","volume":"19","author":"M Javed","year":"2025","unstructured":"Javed, M., Zhang, Z., Dahri, F. H. & Kumar, T. Enhancing multimodal deepfake detection with local-global feature integration and diffusion models. SIViP 19, 400 (2025).","journal-title":"SIViP"},{"key":"61650_CR26","doi-asserted-by":"publisher","DOI":"10.3390\/electronics13152947","author":"M Javed","year":"2024","unstructured":"Javed, M., Zhang, Z., Dahri, F. H. & Laghari, A. A. Real-time deepfake video detection using eye movement analysis with a hybrid deep learning approach. Electronics https:\/\/doi.org\/10.3390\/electronics13152947 (2024).","journal-title":"Electronics"},{"key":"61650_CR27","doi-asserted-by":"publisher","first-page":"4749","DOI":"10.1109\/JIOT.2020.3028574","volume":"8","author":"J Zhang","year":"2021","unstructured":"Zhang, J., Li, Y. & Xiao, W. Integrated multiple kernel learning for device-free localization in cluttered environments using spatiotemporal information. IEEE Internet Things J. 8, 4749\u20134761. https:\/\/doi.org\/10.1109\/JIOT.2020.3028574 (2021).","journal-title":"IEEE Internet Things J."},{"key":"61650_CR28","doi-asserted-by":"publisher","first-page":"24587","DOI":"10.1109\/JIOT.2022.3192322","volume":"9","author":"J Zhang","year":"2022","unstructured":"Zhang, J., Li, Y. & Xiao, W. Toward robust and accurate device-free localization in cluttered environments with commodity wifi devices. IEEE Internet Things J. 9, 24587\u201324599 (2022).","journal-title":"IEEE Internet Things J."},{"key":"61650_CR29","doi-asserted-by":"publisher","first-page":"489","DOI":"10.1016\/j.cmpb.2014.09.005","volume":"117","author":"B Kwolek","year":"2014","unstructured":"Kwolek, B. & Kepski, M. Human fall detection on embedded platform using depth maps and wireless accelerometer. Comput. Methods Programs Biomed. 117, 489\u2013501 (2014).","journal-title":"Comput. Methods Programs Biomed."}],"container-title":["Scientific Reports"],"original-title":[],"language":"en","link":[{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61650-z","content-type":"text\/html","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61650-z.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"text-mining"},{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61650-z.pdf","content-type":"application\/pdf","content-version":"vor","intended-application":"similarity-checking"}],"deposited":{"date-parts":[[2026,7,17]],"date-time":"2026-07-17T13:41:55Z","timestamp":1784295715000},"score":1,"resource":{"primary":{"URL":"https:\/\/www.nature.com\/articles\/s41598-026-61650-z"}},"subtitle":[],"short-title":[],"issued":{"date-parts":[[2026,7,17]]},"references-count":29,"journal-issue":{"issue":"1","published-online":{"date-parts":[[2026,12]]}},"alternative-id":["61650"],"URL":"https:\/\/doi.org\/10.1038\/s41598-026-61650-z","relation":{},"ISSN":["2045-2322"],"issn-type":[{"value":"2045-2322","type":"electronic"}],"subject":[],"published":{"date-parts":[[2026,7,17]]},"assertion":[{"value":"21 May 2026","order":1,"name":"received","label":"Received","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"7 July 2026","order":2,"name":"accepted","label":"Accepted","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"17 July 2026","order":3,"name":"first_online","label":"First Online","group":{"name":"ArticleHistory","label":"Article History"}},{"value":"The authors declare no competing interests.","order":1,"name":"Ethics","label":"Competing interests","group":{"name":"EthicsHeading","label":"Declarations"}},{"value":"All human subject data utilized in this study were obtained from publicly available, de-identified datasets (CASIA Gait Database, Multiple Cameras Fall Dataset, UR Fall Detection Dataset, Recognition of Human Actions, Actions as Space-Time Shapes, HMDB51, and UCF Sports Action Dataset). The original data collections were conducted in accordance with the ethical standards and protocols approved by the respective institutional review boards of the source institutions, and informed consent was obtained from all participants prior to their participation. No identifying personal details, facial features, or original RGB human images are present in the manuscript, and all presented figure elements are fully anonymized.","order":2,"name":"Ethics","label":"Ethical approval and informed consent","group":{"name":"EthicsHeading","label":"Declarations"}}],"article-number":"22535"}}