1. Machine learning-based suspicious activity detection for surveillance application.
- Author
-
Abbas, A. Mohamed, Siddhardha, K., Dattatreya, G. S., and Reddy, K. Srinivas
- Subjects
CONVOLUTIONAL neural networks ,CLOSED-circuit television ,HUMAN behavior ,DEEP learning ,MACHINE learning ,HUMAN activity recognition ,INTRUSION detection systems (Computer security) - Abstract
Among the most significant uses of deep learning techniques in human activity monitoring is suspicious activity identification. Nowadays, people's safety is at the top of the list. This worry stems from the fact that there are more and more danger-inducing actions, from planned acts of violence to injuries sustained in accidents. Closed Circuit Television (CCTV) installations are inadequate because they need the constant presence of a person to oversee the cameras, which is both time-consuming and resource-intensive. A 3D Convolutional Neural Network (CNN)-based Slow Fast Model (SFM) is suggested as part of the study. This model will serve as an automated security system, able to identify unusual behaviours in real-time and immediately notify the relevant departments. The system analyses and detects suspicious human behaviour from CCTV footage in real-time using machine learning techniques. If aberrant activity is found, an alert is generated. A dataset including both typical and out-of-the-ordinary activities was used to evaluate the method, and it achieved an accuracy rate of 93%. [ABSTRACT FROM AUTHOR]
- Published
- 2024
- Full Text
- View/download PDF