Paper Key : IRJ************355
Author: Harshdip Patil,Dhruv Rane,Mohiuddin Shaikh,Prof. Prakash Gadekar
Date Published: 16 Nov 2024
Abstract
This project presents the development of an intelligent violence detection system utilizing computer vision and natural language processing (NLP) techniques to enhance real-time security monitoring. The system is designed to analyze video footage and detect violent actions through a combination of machine learning models and image processing algorithms. By implementing Convolutional Neural Networks (CNN) for object recognition and Long Short-Term Memory (LSTM) networks for sequential analysis, the system can identify aggressive behaviours such as punching, kicking, or crowd disturbances. When a violent activity is detected, the system triggers immediate alerts to authorities, enabling rapid response.In addition to computer vision, the project employs NLP to analyze surrounding audio or textual cues, such as crowd noise or alarm sounds, further refining detection accuracy. This approach improves traditional surveillance by enabling automated, context-aware detection of violence in public areas, schools, and transportation hubs. Key challenges addressed in this study include minimizing false positives, ensuring privacy and data security, and optimizing real-time processing. This system has promising applications in enhancing public safety and is adaptable for deployment in high-risk environments where rapid incident response is critical.
DOI Requested