Article
REAL-TIME ANOMALY DETECTION IN CCTV USING MACHINE LEARNING
The increasing demand for intelligent surveillance systems has led to the development of automated solutions capable of detecting abnormal activities in real time. Traditional CCTV monitoring systems rely heavily on human operators, making them inefficient, timeconsuming, and prone to errors due to fatigue and limited attention span. This research proposes a Real-Time Anomaly Detection System in CCTV using Machine Learning and Computer Vision techniques to overcome the limitations of manual monitoring. The system is designed to automatically analyze video streams, identify unusual activities, and generate alerts without human intervention. The proposed system captures live video footage from CCTV cameras and processes it into frames using OpenCV. These frames are analyzed using machine learning algorithms that compare real-time activity with predefined normal behavior patterns. Any deviation from normal behavior is classified as an anomaly. The system is capable of detecting various abnormal events such as unauthorized entry, suspicious movements, theft, and violent activities. Upon detection, an alert notification is generated and stored along with timestamp details in a SQLite database. A web-based dashboard developed using Flask allows users to monitor alerts and system performance efficiently. The system architecture includes modules such as video capture, frame processing, feature extraction, anomaly detection, alert generation, and database management. The integration of these modules ensures real-time performance and high detection accuracy. Experimental results demonstrate that the system achieves significant accuracy while maintaining a low false alarm rate and fast processing speed. The proposed system reduces the burden on human operators and enhances surveillance efficiency in public places such as airports, banks, railway stations, and shopping malls. It provides a scalable and cost-effective solution for modern security challenges. Future enhancements may include deep learning-based models, cloud deployment, and integration with IoT devices for improved performance. Overall, this research contributes to the advancement of intelligent surveillance systems by providing a robust and automated anomaly detection framework.
Full Text Attachment
