ETMamba: An Effective Temporal Model for Video Action Recognition
Electronics · 23.03.2026 · 15 (6) · 1338
DOI: 10.3390/electronics15061338 ↗Kurz gefasst
Fachartikel in der Zeitschrift Electronics, online erschienen am 23.03.2026. Verfasst von Rundong Hong und 9 weiteren Autor:innen. Die Arbeit nennt 44 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: Jilin Provincial Department of Education, Key Laboratory of Modern Agricultural Equipment and Technology (Jiangsu University), Ministry of Education, Development and Reform Commission of Jilin Province.
Abstract (Originalsprache)
Video action recognition faces persistent challenges in balancing accuracy with computational efficiency. While state space models, such as Mamba, have emerged with linear complexity advantages, they exhibit inefficiency in capturing critical spatiotemporal dependencies within video data. To address this core limitation, this paper proposes ETMamba, an enhanced architecture built upon the Mamba baseline. The ETMamba achieve performance breakthroughs via three core innovation modules: (1) the Spatiotemporal Feature Preservation module retains complete original spatiotemporal correlations before data flattening, solving the problem of spatiotemporal feature loss; (2) the Efficient Bidirectional Sharing strategy accurately models bidirectional temporal dependencies, enhancing key temporal dynamic information; and (3) the Spatiotemporal Collaborative Modulation mechanism combines global temporal and local spatial information to achieve collaborative capture of long-short term dependencies and fine-grained features. We conduct experiments on multiple benchmark datasets, achieving recognition accuracies of 88.3%, 74.6%, 75.7%, and 98.1% on Kinetics-400, Something-Something V2, HMDB-51, and Breakfast datasets, respectively, while maintaining low to medium computational complexity.
Förderung
- Jilin Provincial Department of Education
- Key Laboratory of Modern Agricultural Equipment and Technology (Jiangsu University), Ministry of Education
- Development and Reform Commission of Jilin Province
- Key Laboratory of Urban Land Resources Monitoring and Simulation, Ministry of Natural Resources
- Jilin Provincial Science and Technology Development Plan Project
- National Natural Science Foundation of China
Zitieren
APA
Hong, R., Wen, C., Sun, P., Zhang, L., Niu, Z., Shi, Y., Li, C., Li, M., Su, H., & Chen, H. (2026). ETMamba: An Effective Temporal Model for Video Action Recognition. Electronics, 15(6), 1338. https://doi.org/10.3390/electronics15061338
Vancouver
Hong R, Wen C, Sun P, Zhang L, Niu Z, Shi Y, et al. ETMamba: An Effective Temporal Model for Video Action Recognition. Electronics. 2026;15(6):1338. doi:10.3390/electronics15061338
BibTeX
@article{Hong2026,
author = {Hong, Rundong and Wen, Changji and Sun, Patrick and Zhang, Leyao and Niu, Zhuozhen and Shi, Yaqi and Li, Chenshuang and Li, Mingqi and Su, Hengqiang and Chen, Hongbing},
title = {ETMamba: An Effective Temporal Model for Video Action Recognition},
journal = {Electronics},
year = {2026},
volume = {15},
number = {6},
pages = {1338},
doi = {10.3390/electronics15061338}
}
EUVIMED