Real Time Suspicious Activity Detection in Surveillance Camera Using YOLO V12
Summary
No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.
Summarise with VIRAT Free to try with your Science Catalogs account.Abstract
The rapid rise in criminal activity has highlighted the necessity of intelligent surveillance systems that can detect threats proactively and monitor in real time.Typical drawbacks of traditional deep learning models for identifying aberrant behavior include low accuracy, high computational expense, and limited ability to adjust to changing conditions.This study proposes an improved realtime suspicious activity detection framework that combines a Feature Pyramid Network (FPN) with the YOLOv12 architecture in order to overcome these difficulties.In order to accurately detect small and complex objects in a variety of environmental conditions, the FPN enhances YOLOv12's multi-scale feature representation capability.The proposed system effectively identifies, monitors, and analyzes human activities within surveillance areas while ensuring rapid processing and minimized computational demands.An automated reporting module produces comprehensive analytical summaries to aid security personnel in swift decision-making.Experimental evaluations indicate that the proposed YOLOv12 with FPN model surpasses traditional detection methods regarding accuracy, inference speed, and robustness.This framework provides a scalable, resource-efficient, and dependable solution for contemporary surveillance applications in both public and private sectors.
Fields of Study & Subjects
Video Surveillance and Tracking Methods
Advanced Neural Network Applications
Human Pose and Action Recognition
Publication details
- Journal / source
- Advances in intelligent systems research/Advances in Intelligent Systems Research
- Pages
- 769–788
- Publication date
- 1 April 2026
- Language
- EN
- Document type
- Book chapter