EUVIMEDEUVIMED entdecken
Crossref

NIRBHAY: An AI-Powered Multimodal Threat Detection System for Real-Time Surveillance

Rasika Kulkarni, Rushikesh Jagtap, Sanjyot Pawar, Yami Bhagat, Talat Kudsi

International Journal of Advanced Research in Science Communication and Technology · 01.07.2026 · 577

DOI: 10.48175/ijarsct-36880 ↗

Kurz gefasst

Fachartikel in der Zeitschrift International Journal of Advanced Research in Science Communication and Technology, online erschienen am 01.07.2026. Verfasst von Rasika Kulkarni und 4 weiteren Autor:innen. Die Arbeit nennt 13 Literaturangaben.

AbstractBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Traditional CCTV-based surveillance systems suffer from significant limitations, including dependence on continuous human oversight, operator fatigue, delayed incident response, and frequent missed events. With the rise in violent crimes, weapon-related incidents, and unintentional fire outbreaks, there is an urgent need for automated, real-time threat detection systems. This paper introduces NIRBHAY, an intelligent multimodal surveillance framework that integrates three specialized deep learning modules for comprehensive threat detection. The system combines violence detection, weapon recognition, and fire-smoke identification into a unified surveillance pipeline. For violence detection, we employ a hybrid MobileNetV2–BiLSTM architecture, achieving an F1-score of 0.892, accuracy of 0.893, precision of 0.881, and recall of 0.905. Weapon detection utilizes YOLOv8, attaining a mean Average Precision (mAP50) of 0.927, with class-wise precision of 0.904 for handguns and 0.938 for knives. Fire and smoke detection is implemented using an enhanced YOLOv11 model, achieving mAP50 of 0.941, with fire detection accuracy of 0.958 and smoke detection accuracy of 0.924. The integrated system operates at 25–30 frames per second, enabling real-time performance suitable for live CCTV deployment. NIRBHAY demonstrates robust and reliable threat detection across multiple threat categories, making it highly suitable for deployment in safety-critical environments such as public buildings, transportation hubs, and commercial infrastructures. The system's ability to process multiple threat types simultaneously while maintaining real-time performance represents a significant advancement in autonomous surveillance technology.

Zitieren

APA

Kulkarni, R., Jagtap, R., Pawar, S., Bhagat, Y., & Kudsi, T. (2026). NIRBHAY: An AI-Powered Multimodal Threat Detection System for Real-Time Surveillance. International Journal of Advanced Research in Science Communication and Technology, 577. https://doi.org/10.48175/ijarsct-36880

Vancouver

Kulkarni R, Jagtap R, Pawar S, Bhagat Y, Kudsi T. NIRBHAY: An AI-Powered Multimodal Threat Detection System for Real-Time Surveillance. International Journal of Advanced Research in Science Communication and Technology. 2026:577. doi:10.48175/ijarsct-36880

BibTeX
@article{Kulkarni2026,
  author = {Kulkarni, Rasika and Jagtap, Rushikesh and Pawar, Sanjyot and Bhagat, Yami and Kudsi, Talat},
  title = {NIRBHAY: An AI-Powered Multimodal Threat Detection System for Real-Time Surveillance},
  journal = {International Journal of Advanced Research in Science Communication and Technology},
  year = {2026},
  pages = {577},
  doi = {10.48175/ijarsct-36880}
}
Kostenlos starten →