Sentinel Watch: Harnessing AI and ML for Proactive Crime Prediction and Crowd Management
D. Menaga, J Jenefa Angelin, V Pallavi Priyadharshini · 2024
This research aims to revolutionize crime prevention and public safety by integrating advanced machine learning techniques for real-time crime monitoring and analysis. The proposed system integrates location-based crime classification and video-based violence detection, delivering a comprehensive approach to crime recognition. Leveraging Random Forest, Decision Tree, and K-Nearest Neighbours (KNN) algorithms, the proposed system achieves 99.0% accuracy in classifying crime types using location data inputs. Complementing this, Convolutional Neural Networks (CNNs) are utilized for video analysis, distinguishing violent from non-violent content with 93% accuracy. By combining spatial data insights with real-time video analytics, the framework addresses the dual challenge of detecting crime trends and analysing violent incidents. The exceptional accuracy rates underscore the potential of the system to enhance public safety measures, optimize resource allocation, and enable proactive crime prevention strategies.