EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

GSMTNet: Dual-Stream Video Anomaly Detection via Gated Spatio-Temporal Graph and Multi-Scale Temporal Learning

Di Jiang, Huicheng Lai, Guxue Gao, Dan Ma, Liejun Wang

Electronics · 13.03.2026 · 15 (6) · 1200

DOI: 10.3390/electronics15061200 ↗

Kurz gefasst

Fachartikel in der Zeitschrift Electronics, online erschienen am 13.03.2026. Verfasst von Di Jiang und 4 weiteren Autor:innen. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 53 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: National Natural Science Foundation of China Joint Fund Project, Key Research and Development Program of Xinjiang Uygur Autonomous Region.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Video Anomaly Detection aims to identify video segments containing abnormal events. However, detecting anomalies relies more heavily on temporal modeling, particularly when anomalies exhibit only subtle deviations from normal events. However, most existing methods inadequately model the heterogeneity in spatiotemporal relationships, especially the dynamic interactions between human pose and video appearance. To address this, we propose GSMTNet, a dual-stream heterogeneous unsupervised network integrating gated spatio-temporal graph convolution and multi-scale temporal learning. First, we introduce a dynamic graph structure learning module, which leverages gated spatio-temporal graph convolutions with manifold transformations to model latent spatial relationships via human pose graphs. This is coupled with a normalizing flow-based density estimation module to model the probability distribution of normal samples in a latent space. Second, we design a hybrid dilated temporal module that employs multi-scale temporal feature learning to simultaneously capture long- and short-term dependencies, thereby enhancing the separability between normal patterns and potential deviations. Finally, we propose a dual-stream fusion module to hierarchically integrate features learned from pose graphs and raw video sequences, followed by a prediction head that computes anomaly scores from the fused features. Extensive experiments demonstrate state-of-the-art performance, achieving 86.81% AUC on ShanghaiTech and 70.43% on UBnormal, outperforming existing methods in rare anomaly scenarios.

Förderung

  • National Natural Science Foundation of China Joint Fund Project
  • Key Research and Development Program of Xinjiang Uygur Autonomous Region

Zitieren

APA

Jiang, D., Lai, H., Gao, G., Ma, D., & Wang, L. (2026). GSMTNet: Dual-Stream Video Anomaly Detection via Gated Spatio-Temporal Graph and Multi-Scale Temporal Learning. Electronics, 15(6), 1200. https://doi.org/10.3390/electronics15061200

Vancouver

Jiang D, Lai H, Gao G, Ma D, Wang L. GSMTNet: Dual-Stream Video Anomaly Detection via Gated Spatio-Temporal Graph and Multi-Scale Temporal Learning. Electronics. 2026;15(6):1200. doi:10.3390/electronics15061200

BibTeX
@article{Jiang2026,
  author = {Jiang, Di and Lai, Huicheng and Gao, Guxue and Ma, Dan and Wang, Liejun},
  title = {GSMTNet: Dual-Stream Video Anomaly Detection via Gated Spatio-Temporal Graph and Multi-Scale Temporal Learning},
  journal = {Electronics},
  year = {2026},
  volume = {15},
  number = {6},
  pages = {1200},
  doi = {10.3390/electronics15061200}
}
Kostenlos starten →