AI-POWERED AUTOMATED SURVEILLANCE SYSTEM FOR EARLY VIOLENCE AND THREAT DETECTION
DOI:
https://doi.org/10.5281/zenodo.19577200Keywords:
Violence Detection, Deep Learning, YOLO, Convolutional Neural Networks, Surveillance Systems, Computer Vision, CCTV Monitoring.Abstract
Public safety in crowded and open environments has become a major concern, leading to the widespread installation of surveillance cameras for activity monitoring. Despite their presence, manually watching continuous CCTV footage is labor-intensive, inefficient, and often results in slow responses to violent situations. To address this challenge, this project introduces an automated system that detects violent behavior in surveillance videos using deep learning methods. The proposed framework utilizes the YOLO (You Only Look Once) algorithm to locate and identify people within video frames. Once individuals are detected, a Convolutional Neural Network (CNN) examines their movements and interactions to classify the activity as either violent or non-violent. To enhance the reliability and performance of the model, it is trained and evaluated using established datasets, including Real Life Violence Situations (RLVS), RWF2000, and the Hockey Fight Dataset. When a violent event is recognized, the system automatically logs the timestamp and saves the relevant video frame as visual evidence. System performance is measured using standard evaluation metrics such as accuracy, precision, recall, and F1-score.This solution contributes to smart surveillance by enabling rapid detection of violent incidents, allowing authorities to take prompt action and improve overall public safety
Downloads
Published
Issue
Section
License

This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License.













